U
The InternationalCenterforResearch on the
Management of Technology
The Patterns of Interorganizational
Networks in the Development of
Data Communications Technologies
Pek H. Soh 1
Edward B. Roberts 2
February 1998
WP # 173-98
Sloan WP # 4010
1 MIT Sloan School of Management
and National University of Singapore
2 MIT Sloan School of Management
© 1998 Massachusetts Institute of Technology
Sloan School of Management
Massachusetts Institute of Technology
38 Memorial Drive, E56-390
Cambridge, MA 02139-4307
ABSTRACT
Technology alliances have been used increasingly in the last few years in the data
communications industry. The dynamics of these alliances highlight the dilemma (or
strategy) of firms to achieve interoperability with competing product platforms while
attempting to differentiate and promote their own products. Using social network
analysis, we explore and identify the network patterns of technology alliances in the
development of data communications technologies. To attribute meaning to these
network patterns, a taxonomy of innovations based on the framework by Henderson and
Clark (1990) is created with the help of two well-qualified field experts. Based on 150
sampled firms and 319 technology alliances in the data communications industry from
1985-1996, we observe that the characteristics of core-periphery structure can best
describe the patterns of networks in this industry. More specifically, a small number of
firms in the core of the network have jointly developed modular innovations which have
implications for emerging standards, whereas many other firms have come together for
more limited purposes, supporting incremental innovations at the periphery of the
network. Several issues are raised on the correlation between core-periphery structure
and the nature of innovation and the conditions under which generalizability can be
made. Finally, an important managerial implication is that technology alliances are not
merely a choice for sharing costs and risks with partners but are instrumental to the
standardization and adoption process in the market. The history of technology alliances
among competitors within the industry does determine the trends of emerging
technologies.
ACKNOWLEDGEMENTS
We express our deep appreciation to Professors Michael Cusumano and Scott
Stern for their repeated readings of this material and very useful comments on both
research methodology and the overall presentation. We also thank Professors Lotte
Bailyn, Steve Borgatti, Peter Marsden, Pri Shah, Nancy Staudenmayer, and three
anonymous referees from the 1997 Academy of Management Annual Meeting, as well as
participants at the M.I.T. Behavioral and Policy Sciences Doctoral Proseminar, for their
critiques and helpful suggestions. In addition, special thanks to David Clark of the M.I.T.
Lab for Computer Science, Shamik Chakraborty formerly of Digital Technology Inc., and
other anonymous company executives for their useful insights in the data
communications industry. We appreciate the funding assistance provided by the M.I.T.
International Center for Research on the Management of Technology (ICRMOT), the
M.I.T. Center for Innovation in Product Development (CIPD - National Science
Foundation ERC Award: EEC-9529140), and the National University of Singapore
(NUS). All errors and omissions remain our responsibility.
Please direct questions or comments to:
Pek H. Soh
National University of Singapore
Faculty of Business Administration
10 Kent Ridge Crescent
Singapore 119260
Email: phsoh@alum.mit.edu
1.
Introduction
Strategic alliances have been used increasingly in many technology-intensive
industries since the 1980s (Garrette & Dussauge 1995; Hagedoorn 1995; Hergert &
Morris 1988; Mariti & Smiley 1983).
Strategic alliances are often depicted as
cooperative relationships creating values for partners (Harrigan 1987).
According to
recent theoretical and empirical literature, firms establish strategic alliances in order to
integrate external knowledge (Arora & Gambardella 1995; Hamel 1991; Kogut & Zander
1992), reduce technological and market uncertainty (Mitchell & Singh 1992; Roberts
1985), create synergy between complementary technologies (Olleros & Macdonald 1988;
Sinha & Cusumano 1991), and set forth new standards (Axelrod et al. 1995). Hence,
from strategic management viewpoint, strategic alliances have important implications for
technological innovation.
This paper is focused on the patterns and characteristics of alliances in the data
communications industry. It aims to develop an understanding of the industry in which
strategic alliances have been employed to combine complementary knowledge among
potential competitors. According to the knowledgeable critics in this industry, strategic
alliances have effectively influenced the properties of emerging technologies. The study
therefore raises several implications for the evolution of technology and organization,
which may complement the theoretical and empirical insights arising from the existing
literature in technological innovation and strategic management.
A quantitative method that is useful for understanding the patterns of relationships
1
among firms and that has begun to gain acceptance by some researchers interested in
technological innovation is social network analysis (Freeman 1991).
Social network
analysis allows a researcher to identify and to interpret the patterns of relationships
systematically by attributes of actors and/or relationships.'
Since strategic alliances
generally reflect the cooperation rather than competition among firms, the network
methods which identify cohesive subgroups are employed in the study.
Cohesive
subgroups imply that actors within each subgroup have stronger ties than do actors
between any two subgroups.
In the study, we examine only strategic alliances like licensing and joint
development agreements, which had been intended to create new technologies (referred
to here as technology alliances). Business alliances that do not have significant technical
contributions to the partners are excluded.
Examples of business alliances include
supplier and customer agreements, marketing and distribution agreements, product
bundling, OEM and value-added reselling activities.
Standards-setting alliances that
accelerate the implementation of specifications, and mergers and acquisitions that
eliminate the possibility of interfirm relationships will be omitted also.
In order to
interpret the network patterns using alliance data, a taxonomy of data communications
technologies in terms of radical, architectural, modular and incremental innovations
based on the framework by Henderson and Clark (1990) is defined with the help of two
field experts. Both experts are involved in the interpretation of the agreements.
Using 150 firms and 319 technology alliances in the data communications
In social network analysis an actor can be an individual or an organization.
2
industry from 1985-1996, we demonstrate that social network analysis is useful in
exploring the network structures related to emerging technologies. The primary findings
suggest that firms differ in their roles within the industry network. One central coalition
of firms has repeatedly used technology alliances for the development of modular
innovations which have implications on industry standards. In contrast, other firms have
come together for more limited purposes, supporting incremental innovations at the
periphery of the network. The dichotomy of roles can best be described as the coreperiphery structure. This structure appears to be very stable over a 12-year period. The
findings further raise an intriguing question on the relative importance of technologies
emerged from the industry network occupied largely by the incumbents, versus those
from the new entrants outside the network.
The organization of the essay is as follows.
Section 2 briefly outlines the
objectives of studying interorganizational networks. This is followed by a description of
the data communications industry in section 3 and the research method in section 4. Both
sections 5 and 6 present the findings of social network analysis. Section 7 presents the
discussion and the principal conclusions of the findings. Finally, directions for future
research are suggested in section 8.
2.
The Objectives of Studying Interorganizational Networks
Over the last several years a growing body of literature on interorganizational
networks has been published in some top academic journals for business management
3
(see Table 1).2 These studies have explored the patterns of production networks in
various industries, including automobile, biotechnology, computer, and fashion (Barley,
Freeman, & Hybels 1992; Lincoln, Gerlach, & Admajian 1996; Saxenian 1994; Uzzi
1997). However, a brief review of the literature published from 1990 until June 1997
reveals two difficulties related to the study of interorganizational networks. First, the
analyses based on diverse methodologies and definitions of interorganizational networks
have made the comparison of network characteristics across industries difficult (Jones,
Hesterly & Borgatti 1997). Second, the difficulty in collecting time-series network data
has constrained researcher ability to use longitudinal analysis. Hence, researchers across
disciplinary training have begun to seek a more quantitative technique to study networks.
Table 1. Selected Journals with Articles on Interorganizational Networks 3
Journal
Academy of Management Journal
Academy of Management Review
Administrative Science Quarterly
Organization Science
R&D Management
Research Policy
Strategic Management Journal
1990-1997
6
1
5
2
4
19
3
Before 1990
0
4
7
0
0
3
Recently, technology management scholars have called for the use of social
Three excellent volumes of network studies are available: a collection of essays on networks and
organizations edited by Nohria and Eccles (1992), a special issue in Research Policy (1991, vol. 20, no. 5)
dedicated to the topic of "Networks of Innovators", and a recent special research forum on alliances and
networks in the Academy of Management Journal (1997, vol. 40, no. 2). Using ABI database in
Lexis/Nexis, we searched for article titles with the word "networks" in several refereed journals. Many
articles were retrieved but we counted only articles related to networks of firms, as shown in Table 1. The
oldest articles went as far back as 1977 in AMR.
3 Many studies on interorganizational networks can also be found in journals like American Sociological
Review, Annual Review of Sociology, and Californian Management Review. A more thorough search
2
4
network analysis to identify regularity and characteristics of interorganizational networks
in the development of technological systems (DeBresson & Amesse 1991). 4 By using
social network concepts and methods, researchers are able to explore and formulate
hypothetical behaviors of firms participating in industry networks.
The technology
management literature in fact shows that social network analysis has long been employed
in diffusion studies (Abrahamson & Rosenkopf 1997; Burt 1980; Coleman, Katz &
Menzel 1957; Czepiel 1975) and in communications studies related to problem solving
(Allen 1977; Freeman 1978; Stork 1991). However, how interorganizational networks
evolve and how technological development shapes and is shaped by the evolution of
these networks remain among the important questions for technology and strategic
management scholars (Freeman 1991; Rosenkopf & Tushman 1994).
In the last few
years, many scholars have addressed these questions by examining networks from a
social network perspective (Barley, Freeman & Hybels 1992; Clarysse, Debackere &
Dierdonck 1996; Liebeskind, et al. 1996; Powell, Koput & Laurel 1996; Walker, Kogut
& Shan 1997). These studies have so far been done mostly in the biotechnology industry.
Given the usefulness of social network analysis in studying interorganizational
networks, the purpose of this essay is twofold. The first is to identify the patterns of
interorganizational networks driven by technology alliances in the data communications
including the abstracts will be considered for future literature review.
4 More than a decade ago, Aldrich and Whetten ( 1981) urged organization theorists to study
interorganizational relations from a social network perspective. Although social network theory and
methods have been widely applied in social science research, most of the empirical studies were done at the
individual level. Some network studies published in the 1970s and 1980s were based on local communities
(Gray, 1990). Two early review articles on the use of social network methods for organization studies are
Tichy, Tushman and Fombrun (1979) and Fombrun (1982).
5
industry.
The second is to identify and interpret the characteristics of innovations
specific to particular firms or grouping of firms in the industry. An attempt is made to
integrate the findings within the well-cited innovation framework by Henderson and
Clark (1990). The understanding of the data communications industry network that is
driven more by technology rather than science would provide an interesting contrast to
existing studies in the biotechnology industry.
Since longitudinal data on technology
alliances in this industry are accessible from public databases, the study can trace the
evolutionary patterns of networks which would otherwise be difficult using cross-section
data alone.
3.
The Data Communications Industry
3.1
Technologies and Markets
The data communications industry began to amplify the power of the computer in
the late 1970s with a revolutionary technology--local area network (LAN), which
connects multiple devices such as computers, printers and file servers by a physical
medium in order to provide distributed processing capability for the end-user.5 In 1977
Datapoint Corporation implemented the first commercial LAN technology (ARCnet).
Several competing technologies followed immediately and were aggressively promoted
by some dominant firms in the industry. The LAN standard-setting committee IEEE 802
5 Although data communications between mainframes and terminals began in the 1950s, the birth of the
data communications industry is defined herer by the year 1977 in which the first LAN technology was
commercially available in the market. Appendix A provides a synopsis of local area network standards and
technologies.
6
finally approved Xerox's Ethernet and IBM's Token-ring networks as industry standards
in 1981.6
Many early entrants in the industry during the late 1970s were established firms
from other related industries. In 1982 GE targeted the factory automation market with a
LAN product called GEnet that complied with the IEEE 802 LAN standard. By 1985
AT&T was positioned to compete with IBM on the LAN front with the introduction of its
UNIX-based Starlan. By 1991, however, many independent networking vendors finally
started delivering interoperable technologies across multiple networking platforms. 7 For
example, DEC who had been a key player in Ethernet announced its first token ring
products and Apple Computer unveiled its first gateway for system network architecture
(SNA), the proprietary network architecture of IBM.
Since 1992 intranet and internet activities--connectivity between LAN segments
distributed geographically, have become the main thrust of growth and competition in the
data communications industry.
From 1993 to 1996, we see the emergence of mature
standards in networking hardware with various transmission speeds and greater
bandwidth, enabling applications to be built independent of hardware platforms and
network architectures. 8 There are switched Ethernet, Fast Ethernet (either 10OBase-X or
100VG-AnyLAN), switched Fast Ethernet, Gigabit and switched Gigabit Ethernet.
Token-ring comes in 4- and 16-Mbps (megabit per second) options, as well as a switched
6 Ethernet technology was invented by Robert Metcalfe at Xerox in the 1970s. The technology was later
jointly promoted by Xerox, DEC and Intel from 1976 until about 1981.
7 PC Week, February 10, 1992
8 MIS Week, January 1, 1990
7
version. There are also FDDI and switched FDDI, and ATM in 25, 100, 155 and 622
Mbps. 9 Modems, ASDL and ISDN are other technologies for remote access, allowing
users to connect to their corporate computer systems from homes or office branches.' 0
These new technologies are expected to drive the worldwide LAN market from
$20.5 billion in 1996 to $37 billion in 2000 (see Figure la and Figure lb)." In U.S. 80%
of the companies with more than 100 employees have a LAN and 85% of the employees
in those companies have PCs connected to the LAN. 12 However, 90% of LAN users
remain on Ethernet or Token-ring networks and have not migrated to high-speed
technologies such as 100Base-T and ATM. 13
9 FDDI (Fiber Distributed Data Interface) is a high-speed token-ring networking technology designed to
run over optical fiber at the rate of 100 megabit per second. ATM (Asynchronous Transfer Mode), another
high-speed technology that is designed to be a fast, general purpose transfer mode for multimedia
transmissions.
10 ADSL (Asymmetric Digital Subscriber Line) transmits megabits over twisted-pair telephone lines that
already exist. ISDN (Integrated Services Digital Network) is an older technology that provides greater
speed and bandwidth over regular telephone lines.
1 Doyle Lee, "Global LAN Market: Key Trends and Influences, 1996 to 2000," International Data
Corporation (IDC) Report, No. 12715, December 1996.
12 Ibid.
13Corporate users can deal with the high demand for internet bandwidth by segmenting their networks via
LAN switching technology and installing high-speed LAN backbones.
8
Figure la. Worldwide LAN Market Revenues by Segment, 1996
Remote access
.---. 12%
A]
2
LAN Switches
7%
NICs
19%
Hubs
26%
Routers
24%
1996 Market Size = $20.5 billion
Source: IDC Report, see footnote 10.
Figure lb. Worldwide LAN Market Revenues by Segment, 2000
Other*
2%
Reemote
27
A
Ct
ho, Jwlvllra
22%
NICs
9%
AT
6%
\
Hubs
Hubs
17%
Routers
17%
2000 Market Size = $37 billion
*Other includes Gigabit Ethernet and other new technologies.
Source: IDC Report, see footnote 10.
3.2
Data Communications Firms
We compiled the list of firms in the U.S. data communications industry using the
CorpTech directory which has been published yearly since 1986.14
The CorpTech
directory consists of the profiles of all public and private firms involved in technology-
9
intensive industries by product types. Figure 2 below shows the distribution of firms
founded in the industry between 1980 and 1996. The number of new entrants during this
period is 362. The number of firms founded before 1980 in other related industries like
computer and telecommunications approximately accounted for another 134 entrants.
Figure 2. Number of Entries and Exits by Year
in the Data Communications Industry
300
250
O 200
E 150
Z 100
50
0
- Entries
-
Acquisitions/Exits
I
-A-Total in industry
o
00
N
00
t
00
\D
00
00
00
O
0\
"
CO\
It
0
O\
Year
Source: CorpTech Directory (1986-1997) and Lexis/Nexis (1997)
While collecting information from public sources, we also found that the number
of mergers and acquisitions was increasing in the 1990s. The number of acquisitions per
year is included in Figure 2. The total number of firms in the industry by 1996 is 496 and
the total number of acquisitions is 168, almost 34% of the firms being absorbed within
the industry. 5
Besides acquisitions, dissolution and bankruptcies were checked using the
14 The
CorpTech Directory, Corporate Technology Information Services, Inc. Woburn, MA.
Internet:http//www.corptech.com.
15 Based on recent industry reports in Dow Jones Wires, the number of acquisitions is rising in this
industry. The actual number of acquisitions could be higher since many small acquisitions are not widely
reported in news announcements.
10
Directory of Obsolete Securities and the bankruptcy database in Lexis/Nexis.16 Only two
bankruptcies were detected.
search.
Two reasons may account for the low outcome of this
First, information on private firms is not well covered in the bankruptcy
databases and there are about 250 private firms in the industry. Second, many firms that
entered the industry but left within the first three years were not published in the
CorpTech directory in the first place. Therefore, the number of exits in Figure 2 basically
reflects the number of acquisitions in the industry.
4.
Research Method
4.1
Social Network Analysis
Social network analysis is focused on uncovering the relational patterns of
interaction among individuals or organizations.
Most social network analyses can be
simplified into three steps. The first is to generate some structural patterns from the
relational data. Then the characteristics of these patterns are identified and interpreted
using additional information such as attributes of actors and their relationships.
t7
Finally, implications are drawn as to how the specific patterns and their meaning advance
our insights into the relevant field of research. The following sections explain in details
the methods applied in the first and second steps. The findings and their implications are
discussed in Sections 5 and 6.
Directory of Obsolete Securities, 1991 Edition, Financial Information, Inc. Jersey City, NJ.
17 Actors are referred to as individuals or organizations.
16
11
Step 1: Representation of Network Structures
Two general techniques can be applied to represent network structures, i.e.
relational and positional methods (Dimaggio 1986; Mizruchi & Galaskiewicz 1993;
Wasserman & Faust 1994). Relational methods identify cohesive subgroups within a
network based on specific properties of ties between actors.
Depending upon the
hypothetical social processes, the properties of ties impose certain restrictions on the
network structures. For example, we can formulate the patterns of information diffusion
through intermediaries by specifying the number of intermediate ties between any two
actors. In contrast, positional methods partition actors based on the similarity of their ties
to all other actors and stress competition among actors within a structural position. Two
actors are said to be "structurally equivalent" if they have identical ties to and from all
other actors in the same network. These two actors may or may not have direct ties with
each other but they are competing for resources provided by their identical partners.
Since technology alliances in the data communications industry reflect the
development of highly inter-dependent technologies, relational methods are conceptually
more sensitive to identifying clusters of cooperative relationships.
Furthermore, the
resulting cohesive subgroups can be interpreted in the second step by the substantive
contents of their partnership agreements.
Various graph theoretic techniques are
available to identify cohesive subgroups but many of them require the properties of ties to
be specified beforehand (Wasserman & Faust 1994). Since the organizing concepts and
underlying dimension of the industry network is not known a priori, we employ a less
12
restricted approach, i.e. multi-dimensional scaling (MDS), to display the proximity of
firms in the network graphically.
The use of MDS to represent the relative distance and closeness of actors within a
network is typical of social network analysis (Coxon 1982; Schiffman, Reynolds &
Young 1982). Based on some measure of pairwise proximity among actors, such as the
geodesic distance between each pair of actors, MDS will compute a set of estimated
coordinates among pairs of actors.' 8 In a two-dimensional spatial map, MDS represents
"similar firms" as coordinates close to each other whereas "dissimilar firms" as
coordinates distant from each another. Tables B-1, B-2 and B-3 provide examples of the
output in generating the MDS map by UCINET, a software program for social network
analysis. 19
Step 2: Interpretation of Network Structures
The interpretation of an MDS map is twofold.
First, we look for significant
patterns in the MDS configurations, i.e. detect structures that are not simply arbitrary
artifacts of the scaling procedure but are stable characteristics of the original data. Then
we ascribe a meaning to these structures using additional information such as properties
of firms and their relationships with others.
18 In graph theory, a geodesic is a shortest path between two nodes. There could be more than one shortest
path between two nodes. The geodesic distance is defined as the length of a geodesic between the two
nodes.
19 Borgatti, Everett and Freeman, 1992, UCINET IV, Columbia: Analytic Technologies. Also special
thanks to Professor Borgatti for his help and suggestions in the use of the software and social network
analysis.
13
The graphical output of MDS is depicted in this study as a set of firms connected
by lines, indicating the presence of alliances between two firms. 20 In addition, thicker
lines will be used for any two firms that have relatively more alliances with each other in
the network. Three aspects of MDS configurations that have received much attention are:
1) the dimensions (orthogonal axes which span the MDS space), 2) the simple graphical
structures (chains, circular and etc.), and 3) the regions (high density of points relatively
to low density of points) (Coxon 1982). In the study, regions with varying degrees of
density are expected to emerge since firms with more alliances among themselves are
likely to cluster in space. For this reason, a network measure called centrality, which
highlights the structural differences between firms by their relative locations in the
network, would be useful in simplifying the network patterns.
Three different measures of centrality are commonly used in social network
analysis, namely, degree, betweenness and closeness.
Among the three measures,
closeness centrality is more appropriate because it considers how central the firm is with
respect to all others in the same network. On the other hand, degree centrality counts the
number of direct ties from the focal firm only and betweenness centrality assumes that
the importance of a firm is reflected in the firm's potential control over the paths
connecting two other firms. Appendix C provides additional information on closeness
centrality.
Finally, the contents of technology alliance agreements provide a substantive
The graphics are generated by the software called Krackplot 3.0, 1995, Krackhardt, D. Blythe, J., &
McGrath C., Natick, MA: Analytic Technologies. http://analytictech.com/
20
14
interpretation of the network patterns. A conceptual framework defining the nature of
innovations is thus necessary for systematic analysis of the contents. The notion that
different types of innovations exist has important implications in the studies of
technological innovation. In particular, following Schumpeter's (1942) notion of radical
and incremental innovations, studies have found evidence of radical innovations
destroying existing capabilities and markets whereas incremental innovations enhance
existing skills and knowledge (Abernathy & Clark 1985; Christensen 1995; Henderson &
Clark 1990; Tushman & Anderson 1986).
The following section describes how a
taxonomy of innovations in the data communications industry was designed for this
study.
4.2
Taxonomy of Innovations
In this study technology alliances involve technological developments in the
network environment of the International Standards Organization (ISO) reference model
as shown in Figures A-2 and A-3 of Appendix A. These developments commonly reflect
a need for hierarchical connectivity among the network components.
Each ISO layer
defines a particular protocol necessary for data transmission with specific capability.
Interconnectivity across layers further ensures that data are translated into appropriate
codes handled by adjacent layers. Therefore, technological innovations happen within
and across layers in the network environment. A taxonomy of innovations that defines
the extent of design changes made in the network environment would be useful for
15
classifying the technology agreements.
Owing to the complexity of networking technologies, an engineer who has been
developing products for the network environment for the past three years was recruited to
help define the taxonomy. First, he was briefed on the purpose of his tasks and given the
literature by Henderson and Clark (1990), Christensen (1992a, b), and Tushman and
Anderson (1986) to review the various theoretical frameworks. We then examined the
publications available in the International Data Corporation's (IDC) library to explore
other possible models employed by industry analysts. Not surprisingly, the IDC industry
reports tend to focus on general product features from the end-user's viewpoint. We felt
that a conceptual model which addresses the technical implications of interfaces and
components separately would be more useful for our purposes.
Given that technical changes involve both ISO layers and their linkages, the
framework of component and architectural innovations developed by Henderson and
Clark (1990) appears to be applicable, as shown in Figure 3. Four types of innovations
are defined: incremental, modular, architectural and radical. The collaborating engineer
felt that radical innovations which change the core concepts and redefine the linkages
rarely occur in this industry. The first radical innovation in the data communications
industry was probably the concept of local area networking itself. In his opinion no
technological development in the past 12 years from 1985-1996 can be qualified as a
radical innovation.
16
Figure 3. A Taxonomy of Innovation Adapted from Henderson and Clark (1990)
Core Concepts of ISO Layers
Reinforced
Overturned
Reinforced
Incremental
Innovation
Modular
Innovation
Changed
Architectural
Innovation
Radical
Innovation
Linkages
between
ISO Layers
Architectural innovation is rearrangement of the way in which component layers
relate to each other. The basic purpose of each layer remains unaffected. However, new
functionality may be added into the layer in order to support the rearrangement.
example, Ipsilon Networks'
architectural innovation.21
IP (Internet
Protocol) switching technology
For
is an
Ipsilon Networks claims that the technology resolves the
problems of complexity, inefficiency and functionality duplication inherent to the
traditional networking approaches.
In essence, IP switching goes against the conventional layered architecture of neighbor-layer opacity and
ties up the IP Layer with the ATM (Asynchronous Transfer Mode) Adaptation Layer by defining two new
protocols called GSMP (General Switch Management Protocol) and IFMP (Ipsilon Flow Management
Protocol). The basic approach is to route IP traffic (Layer 3) on top of ATM Adaptation Layer (Layer 2).
The traditional approaches to this end have been LANE (LAN Emulation) defined by ATM Forum,
Classical IP over ATM from the IPATM working group of IETF and MPOA (Multiprotocol over ATM)
under ballot from ATM Forum.
21
17
Modular innovation involves a replacement of protocol or technology for a given
ISO layer. This can be viewed as a fundamental change in the technology used in that
layer, leaving the other layers unchanged. The entry of ATM (Asynchronous Transfer
Mode) technology in early 1990s in the networking industry is an excellent example of
modular innovation. 22
Finally, incremental innovation adds new functionality to existing technologies
for one ISO layer and enhances the interface with adjacent layers which use different
technologies. Gigabit Ethernet is a good example of incremental innovation.
Gigabit
Ethernet technology fully capitalizes on the existing installed base of Ethernet technology
(IEEE 802.3 CSMA/CD Layer 2 Protocol). The basic limitation of traditional Ethernet
technology was limited bandwidth (10 Mbps).
Gigabit Ethernet technology uses the
same Layer 2 Ethernet technology over Fibre Channel Physical Layer (Layer 1) to
achieve bandwidths of the order of 1 Gbps.
The coding of technology alliances based on the above taxonomy of innovation
remains a difficult task since a good understanding of the history of technological
developments and the technical details is necessary. Although a panel of industry experts
would be ideal, owing to the time constraint to complete the study, another expert (Expert
2, E2) was sought to perform the coding in parallel with the first engineer (Expert 1, El).
The main advantage of having independent experts to apply the taxonomy is to resolve
ATM is basically a Layer 2 technology. ATM delivers important advantages over existing LAN and
WAN technologies, including the promise of scaleable bandwidths at unprecedented price and performance
points and Quality of Service (QoS) guarantees, which facilitate new classes of applications such as
multimedia.
22
18
any problematic issues arising from coding. Comparison of coding outcomes is made to
ensure reliability of scores. Through an interview with a founder and chairman of a
prominent networking company in Massachusetts, E2 was identified and recruited to
assist. This second expert is most qualified, a Senior Research Scientist who has been
developing advanced communications techniques in the M.I.T. Laboratory for Computer
Science and is well-known in the industry.
He has been actively involved in the
development of local area networking technologies since the 1970s and continues to
attend many standards meetings and professional conferences in the industry.
In
addition, he serves as a board member of a networking company.
E2 was briefed on the purpose of his tasks and given the taxonomy of innovations
to examine.
He did not find particular problems with the various definitions in the
taxonomy. He was later told that El had assisted in coming up with the definitions and
would be coding the alliance agreements independently. Both experts were given Tables
D- 1, D-2 and D-3 in Appendix D and a total of 61 alliance agreements to code.
4.3
Longitudinal Network Data
Technology alliances are used in this study to represent the formal relationships
between firms.
For each of the 496 firms found previously, full-text descriptions of
technology alliance agreements, if any, were retrieved from Lexis/Nexis and Dow Jones
Wires.
Most agreements included joint product development, licensing and/or cross-
licensing agreements.
Figure 4 below shows the distribution of alliance agreements
19
established in the U.S. from 1980 to 1996. There are 613 joint product development
agreements and 157 licensing agreements. In fact, the total number of agreements should
be higher since we omitted agreements that were not publicized, international agreements
involving firms outside of U.S., and minority equity investments. 2 3
Figure 4. Number of Joint Product Dewlopment
and Licensing Agreements by Year in
the Data Communications Industry
150
1oo
0
1;. %
I
\ I
t~v.
-n-
LI
0
o
00
O
G
00
\ O\'
t
00
0'
10
00
0\ O\
00
00
0
O
O\
N
08
O
O
0
\
a\
0'
Year
Source: Full-text reports from Lexis/Nexis and Dow Jones Wires, 1997
Figure 4 also shows a downtrend in the alliance activity from about 1995. Three
factors may have contributed to this decline: 1) the decrease in the number of new
entrants, 2) the increase in the number of acquisitions, and 3) data not yet available online in the public databases.
Technology alliances were sparse before 1985 because early technological
developments were largely focused on proprietary technologies. For this reason, only
Minority equity investments have been used as an incentive to motivate the invested firms to undertake
product development projects that meet the objectives of the investing firms. However, minority equity
investment data were not widely reported in public sources for the intended analyses later.
23
20
alliance agreements from 1985 to 1996 were examined. All input data to UCINET are in
the form of NxN matrices, where N is the number of firms involved in technology
alliances and each cell indicates the number of agreements between two firms.
More than 150 firms were detected over the 12-year period.
Given the large
matrix, the graphical representations of MDS would be extremely difficult for the
analyses. Consequently, we split the dataset into three time periods that approximately
coincide with the emergence of particular industry standards: 1985-1988, 1989-1992, and
1993-1996.
According to various industry sources, 1985 marked the year of market
growth with two competing standards, Ethernet and Token Ring, dominating the
development of new component technologies. In 1989 one of the first industry standardssetting forums was initiated by competing firms to advance the FDDI STM standard (an
improved protocol over token ring networks.) Since then, many large standards-setting
forums have been promoted by major industry players rather than the standards-setting
bodies like IEEE, ANSI and ITU (see Appendix F for standards-setting forums.) 24
Finally, 1993-1996 represents the dramatic growth of data communications
markets driven by internet and intranet activities. About half of the total technology
alliances were established during this period.
Although the division into these time
periods seems somewhat arbitrary, subsequent analyses based on the variation in the
global network patterns will confirm the significance of industry forces such as market
growth and standardization in driving technological changes over time.
IEEE is the Institute of Electrical and Electronics Engineers, Inc., ANSI is American National Standards
Institute, and ITU is International Telecommunications Union.
24
21
For the three periods, there are 32, 127 and 211 firms, and 36, 118 and 207
agreements respectively.
However, some of these alliances were non-repeated, dyadic
relations between two firms that are isolated from all other firms. Since isolated firms
cannot be reached by others, closeness centrality measure will assign zeros to all firms in
the network. For this reason, isolated relationships were omitted in the analyses. In the
final sample, we ended up with 28, 68 and 117 firms and 34, 89 and 196 agreements in
the corresponding periods. Table 2 below gives the distribution of the most recently
founded firms by period. Overall, few firms in the sample are less than 8 years old.
Table 2. Number of Most Recently Founded Firms in the Sample by Period
Period
1985-1988
1989-1992
1993-1996
Percentages
5.
Total no. of firms in
No. of sample firms in
No. of sample firms
the industry founded
this period
founded in this period
in this period
113
28
5 (18%)
82
68
5 (7%)
43
117
4 (3%)
in parentheses are the proportions of sample firms by period.
No. of sample firms
founded in the 4 years
prior to this period
7 (25%)
17 (25%)
10 (8%)
The Characteristics of Network Structures and Memberships
The patterns of the industry networks in three periods are presented in Figures 5,
6 and 7. Each line joining any two firms represents the presence of alliances but a thicker
line indicates 2 or more alliance agreements were established in separate years. Since the
25 The total number of alliance agreements from 1985 to 1996 included in the matrices is 361, which is
slightly more than half of the reported number 684 in Figure 3. Two reasons account for the discrepancy.
First, those omitted agreements involved largely private firms and some non U.S. parent firms for which we
could not obtain sufficient information. We decided to drop these firms in the study. Second, Figure 3 has
incorporated the latest results from the second and third searches through Dow Jones Wires, which
produced many new agreements not found previously in Lexis/Nexis. Nevertheless, the set of 361 alliances
22
list of alliance agreements for each period was compiled from public sources, some
agreements might have been omitted in the analysis. Despite the omissions, the global
representation of the industry networks can be revealing in regard to emerging network
characteristics over time.
In Figure 5, the network pattern from 1985-1988 shows three branches
intersecting at 3Com (3CO) and Microsoft (MS) in the center of network. Both firms
were known to be developing networking software and hardware, such as OS/2 LAN
Manager, compatible to the IBM PC platform during this period. Several other notable
firms like DEC, AT&T, Intel (INT), and Novell (NVL) occupy positions at the three
branches respectively.
DEC was developing various technologies including modules
connected to baseband Ethernet, broadband Ethernet, and WANs. Novell was
implementing connectivity for PC workstations to LAN servers through TCP/IP gateway,
whereas Intel was developing Ethernet and ISDN chips for Ethernet and remote access
technologies.
has at least incorporated the prominent agreements in the industry during the last 12 years.
23
Figure 5. Network of Firms in the Data Communications Industry, 1985-1988
SMC
MT
INT
\/
ATT
RT
SC)
v t
NET
/ET
I M
\G | j
O0---- B C
DG
RAB
C]
DCA
ORV
STR
DSC
SNP
WSD
24
Figure 6. Network of Firms in the Data Communications Industry, 1989-1992
SIL WDT
LSI
FOR
FUJ
CY
1
ULT
/
LUX
/
TRW
XCO--GDC
-- GDL
COD
TT
ASC
IDE
INT
VLK
I
XPL
NLK
From Figure 5 to Figure 6, however, four significant characteristics of
interorganizational dynamics emerged over time. First, the industry network evidently
has become denser and more connected among firms. Second, some firms like DEC, HP,
Novell (NVL), and SynOptics Communications (SNP) from the branches in the last
period have moved closer to the center of the network by forming relationships with other
25
existing firms, especially 3Com that has continued to maintain the central position. In
contrast, others like AT&T, Intel (INT) and UB Networks (UBN) have remained in their
branch positions. Third, repeated alliances are apparent among some firms from the last
period. Finally, new branches have emerged, led by Cabletron Systems (CBT), Cisco
(CSO) and Wellfleet Communications (WFL) that did not exist in network before.
Figure 7. Network of Firms in the Data Communications Industry, 1993-1996
g l.
26
Interestingly, Figure 7 reveals some similar network characteristics in the third
period. First, even more firms participate in the industry network. Second, 3Com, HP,
IBM and SynOptics (SNP) have maintained their central positions by forming more
alliances.
The exceptions are DEC and Novell (NVL) which appear to have fewer
alliances compared to the last period.
Third, companies like AT&T, Cisco (CSO),
Ericssons (ERC), Intel (INT), NorTel (NT, formerly known as Northern Telecom), and
UB Networks (UBN) have moved to the center of the network from their boundary
positions, whereas Wellfleet (WFL) and Cabletron have not. Finally, many new branches
that are relatively well connected with others along the boundary have emerged. These
positions are led by General Datacomm (GDC), Optical Data Systems (ODS), and Sun
Microsystems (SUN), and so on.
Further investigation suggests that no obvious firm characteristics like founding
age and firm size can be associated with the dynamics of interorganizational
relationships.
However, some network characteristics like centrality can be useful in
summarizing the patterns of network structures. Table 3 shows the characteristics of
networks in three periods by closeness centrality and network centralization measures. In
each successive period, the size of the network (number of firms) is twice as large and the
number of alliances is more than double.
The statistical means of the firm-level
closeness centrality are almost the same and the standard deviations are low across the
three periods.
The network centralization index is a group-level measure of variation in
The normalized closeness centrality indices are expressed in percentages by UCINET. Firm-level
indices are not shown here except for the statistical mean.
26
27
firm-level centrality. The larger the index the more likely that a particular firm is quite
central, with remaining firms considerably less central.
The network centralization
indices here are relatively small and stable (26.83-29.97%).
Table 3. Characteristics of Networks by Closeness Centrality and Network Centralization
Closeness Centrality
Mean at firm-level
Std Dev
Minimum
Maximum
Network Centralization
Number of Firms
Number of Alliances
1985-1988
26.12%
5.74%
17.76%
40.30%
29.97%
28
34
1989-1992
23.69%
5.04%
14.50%
36.81%
26.83%
68
89
1993-1996
25.87%
4.88%
14.85%
39.32%
27.26%
117
196
The high stability of aggregate network properties does not, however, imply a low
turnover of network memberships across time (Morgan, Neal & Carder 1996).
The
distribution of firms participating in the networks is given in Table 4. Only 14 firms
(9%) continued to form alliances in all three periods. 33 (22%) additional firms formed
alliances in two consecutive periods and 101 (67%) firms participated in a single period
only. However, these numbers may be different since 72 firms are right censored in the
third period, i.e. no data available as to possible firm activity later than 1996. Overall, we
observe a low stability of network memberships.
28
Table 4. Characteristics of Network Memberships
Period(s)
Number of firms*
The first period only (1985-1988)
8
The second period only (1989-1992)
21
The third period only (1993-1996)
72
The first and second only
4
The first and third only
2
The second and third only
29
All three periods
14
Total
150
* The number of firms participating in specific period(s) only.
Given that the exploitation of external sources of innovation is not limited to
alliances, the stability of network memberships can also be affected by the mergers and
acquisitions activities in the entire industry. In terms of the relative use of alliance and
acquisition strategies from 1985-1996, 8% (42 out of 496) of the firms used both alliance
and acquisitions, about 22% (108 out of 496) used alliances alone and 17% (85 out of
496) used acquisitions alone.27 Slightly less than 50% of all firms in the industry use
either or both alliances and acquisitions.
Table 5 indicates that only 18 (11 %) firms participating in the networks had been
acquired. Of the total 160 acquisitions, 90 (56%) were made by non-members of the
networks whereas 70 (44%) were made by 42 members of the networks. From 1989 to
1996, however, the acquisition trend of members (columns 3 and 5 in Table 5) is rising
faster than that of non-members (columns 4 and 6 in Table 5), presumably reflecting the
larger size of network members.
Note that all the percentages reported in this paragraph could be higher since the total number of 496
firms has not accounted for the number of acquisitions each year.
27
29
Table 5. Number of Acquisitions Made by Members and Non-members of Networks
Period
1985-1988
1989-1992
1993-1996
Total
Total no. of firms
acquired in the
period
No. of non-members
acquired by
members
No. of non-members
acquired by nonmembers
No. of members
acquired by
members
No. of members
acquired by nonmembers
22
43
95
160
4
10
42
57
15
31
39
85
1
2
11
13
2
0
3
5
The above results suggest that instead of being an acquisition target themselves,
members of the networks have become more likely to acquire non-members.
phenomenon raises two interesting empirical questions.
This
Are non-members of the
networks more likely to have better technologies which lead to their acquisitions by
others? Are network members more likely to maintain their status quo owing to their
prior relationships with others?
To examine these questions systematically, more
empirical data on non-members as well as the relative motives of acquisitions and
alliances are required.
More implications of the questions will be discussed in the
conclusions of this essay.
To further understand the dynamics of networks, Table 6 below lists the identities
of only "high-centrality firms" that have closeness centrality one standard deviation
above the statistical means. All 19 firms in the table are early entrants in the industry,
even really including Bay Networks, established in 1994 as a result of a merger between
SynOptics (1985) and Wellfleet Communications (1986).28
In Table 6, the founding year of the firm is stated if it was a new entrant in the data communications
industry. In the case of established firms from related industries like computers and telecommunications,
we stated the year in which their networking divisions were formed (like AT&T, HP, Ericsson and
28
30
I
Table 6. Subset of High-centrality Firms
-
1985-1988 (4 firms)
1979 3CO* (3COM)
1983 HP*
1984 MS " ' (Microsoft)
1979 TND* (UB
Networks)
I,~
1989-1992 (11 firms)
1979 3CO* (3COM)
1976 DEC2
1982 EXC"' (Excelan)
1983 HP*
1979 IBM*
1984 MS 1'2 (Microsoft)
1983 NET 23' (Network Equipment
Technologies)
l'~
1973
NT 2 3 (NorTel)
1983 NVL* (Novell)
1985 RTX' 2 (Retix)
1985
SNP2 3 (SynOptics)
_
1993-1996 (15 firms) __________
1979 3CO* (3COM)
1983 ATT*
1994 BAY (Bay Networks)
1983 CHP2 3 (Chipcom)
1984 CS 2 '3 (Cisco)
1985 ERC 2 3 (Ericsson)
1983 HP*
1979 IBM*
1976
INT3 (Intel)
1977 MOT3 (Motorola)
1973 NT2 3 (NorTel)
1983
NET2 '3 (Network Equipment
Technologies)
1985 SNP2 3 (SynOptics)
1986 STR3 (Stratacom)
1979 TND* (UB Networks)
14% (4 out of 28 firms)
16% (11 out of 68 firms)
13% (15 out of 117 firms)
Note: High-centrality firms here have indices one standard deviation above the corresponding means in
Table 3. Year of founding the networking business is given beside each firm. The superscript number
indicates the period in which the firm has centrality above the mean. An asterisk shows that the firm has
centrality above the mean in all three periods. The bottom row indicates the percentage of high-centrality
firms in each network.
15 firms (including Bay Networks in its former identity as SynOptics) were active
in at least two periods. 13 firms have made a total of 35 acquisitions, that is 50% of the
acquisitions made by all network members over three time periods (see also Table 5).
Among these 13 firms, only 3Com, Cisco and Novell have acquired other high-centrality
firms, namely, Chipcom, Stratacom and Excelan respectively. The two latter acquired
firms had been partners of the acquiring firms prior to acquisitions.2 9 Independent sales
Motorola) or some of their first projects related to LAN were announced (like DEC, IBM, Intel and
Microsoft).
29 Among all acquisitions in the industry, a total of nine firms have acquired their former alliance partners.
These firms are Bay Networks, Cisco, General Instrument, Network General Corp., Network System Corp.,
Networth, Novell, Telco Systems and Synernetics. Bay Networks, Cisco and Novell are the only highcentrality firms.
31
data from 1993 to 1996 were not available for Chipcom, Stratacom, SynOptics, UB
Networks and Ericsson.
Of the 10 other firms, only NET is not among the top 50
producers in 1994 and 1995 sales revenues ranked by NetworkWorld. 30
High-centrality firms can be broadly characterized as early entrants and market
share leaders in the industry. They are also more likely to make acquisitions compared to
low-centrality firms. 3 The above observations lead to an interesting question: do firms
with more resources also attract more partners? If they do have more resources than
others, what then are the benefits of forming more alliances?
The technology
management literature has informed us that larger, established firms tend to be less
innovative than new entrants. Consequently, alliances and acquisitions may be the best
alternatives for these larger firms to gather new ideas so as to complement their internal
capabilities.
However, if alliances bring external sources of innovation, why then are not more
firms like those with low-centrality forming more alliances? Recall that low-centrality
firms are also less likely to make acquisitions. Low-centrality firms either do not have
the resources to attract potential partners or they are more innovative in the first place. It
is apparent that the factors underlying the choice for alliances, acquisitions and internal
development differ between high-centrality and low-centrality firms.
All the above
questions thus have compelling implications for the evolution of technology.
More
NetworkWorld, December 30, 1996. The top 50 firms include many established firms from the
telecommunications and computer industries. These firms are involved in other areas of data
communications technologies such as software and wireless technologies that are not included in the study.
31 Low-centrality firms have centrality below the means, whereas others with centrality equals to or above
30
32
empirical data related to technology alliances will be analyzed in the next section in order
to understand the potential roles played by high-centrality versus low-centrality firms in
the industry network.
Despite their differences, high-centrality firms and low-centrality firms are two
potential sources of stability in the global network structure. The first source is a set of
"core" ties that are relatively stable and have anchored the composition of the network
over time.
The second source is a set of "peripheral" ties that are relatively
interchangeable, rendering the aggregate network properties unchanged.
These two
sources are not mutually exclusive, so either or both could be a source of network
stability.
Three explanations are possible for the above speculation. First, a small number
of firms might proactively be using alliances as a strategy for a variety of reasons,
acquiring complementary
technologies and expertise, speeding up new product
development and monitoring emerging technologies. In contrast, most firms are probably
using alliances as the best option at the time relative to in-house development. While the
benefits of strategic alliances are manifold (Contractor & Lorange 1988; Hagedoorn
1993; Tucci & Cusumano 1994), many firms are still skeptical about forming
partnerships with present or potential competitors.
Asymmetric attributes of the firms
can change the relative bargaining power as well as the relative competitive standing
between partners (Hamel 1991; Yan & Gray 1994).
Even when such skeptical firms
participate in joint projects, these alliances are less likely to be involved in highly
the means will be high-centrality firms.
33
sensitive areas of technological innovation.
Second, from the social capital viewpoint (Burt 1992b; Coleman 1990), firms that
are more "central" in the network by virtue of maintaining many direct and indirect
relationships with all other firms have better access to information leading to potential
alliance partners. On the other hand, firms located on the periphery of the network obtain
lower information benefits, hence have lower inclination for new relations.
Walker,
Kogut and Shan (1997) showed that the development and maintenance of social capital
have led incumbent firms to reproduce the existing network structure, giving rise to
network stability. Similarly, Gulati (1995) found that the social structure of past alliances
explains the formation of new interorganizational relationships. The stability of highcentrality firms versus the instability of low-centrality firms in this study is good
evidence of the varying opportunities and constraints in information access available to
the two types of firms.
Third, with increasing networking experiences, the marginal costs for central
firms to form new alliance relations can be assumed to be decreasing. However, the
marginal costs are likely to be relatively higher for the boundary firms.
From the
transaction costs viewpoint, firms are assumed to be opportunistic and the market for
technological knowledge is imperfect, thus leading to high contractual costs in
technology exchange transactions (Teece 1992; Williamson 1985). On the other hand,
reciprocity and trust are important social factors in any bilateral transactions (Bradach &
Eccles 1989; Ouchi 1980; Ring & Van de Ven 1992). Richardson (1972) pointed out that
34
firms achieve stability and reliability in their relationships with other firms through
qualitative coordination. Closeness of ties, reciprocity and trust can all play a part as a
cost-effective monitoring device for firms' deviant behaviors. The ability of central firms
to manage a diversity of new and repeated relationships with particular partners in the
past certainly reflects the potential trust level in new transactions, hence sustaining the
stability of their participation in the network.
The relatively few new entrants that
participated in the network, as shown in Table 2, may in fact indicate that young
organizations require time to build up their reputations and perhaps also their internal
capacities necessary to be in the network.
Above all, resource dependency, information benefits and networking experience
are potential explanations that support the stability of "core-periphery" structure of the
industry network.
In the following section, we explore the correlation between core-
periphery structure and nature of emerging technologies by using the coding scores from
two industry experts.
6.
The Link Between Core-Periphery Structure and Nature of Emerging
Technologies
Tables D- 1, D-2 and D-3 in Appendix D list by period the contents of technology
alliances for the subgroups in Table 7. These subgroups were selected randomly from
Figures 5, 6 and 7 respectively.
Ideally, all alliance agreements should be evaluated.
35
Unfortunately, the time available from each expert was limited to at most three hours.3 2
Hence, the interpretation of alliances associated with the core-periphery structure must be
treated with caution.
Table 7.
Selected Cohesive Subgroups by Period
' 1989-1992
Cabletron, Cayman Systems
Bytex, Crosscom
1993-1996
1.
2.
1985-1988
Chipcom, DEC
Codenoll Technology,
1.
2.
3.
Sytek
IBM, Network
3. Excelan, Network System Corp.,
Wellfleet Communications
Equipment
Technologies
4.
5.
Tiara, Sun
Netrix, Pacific Communication
Microsoft, 3COM, UB
Networks
Excelan, Novell, Oracle
6.
7.
Sciences
Ericsson, Luxcom
3COM HP
6.
SynOptics, Western
Digital
8.
9.
Chipcom, IBM
National Semiconductor, Network
7.
AT&T, Intel
Equipment Technolozies
10. Fujitsu Network Systems, Tellabs
11. Codenoll Technology, Lanex
12. Madge Networks, Cisco,
SynOlptics
13.
14.
AT&T,HP
Cascade Communications, Motorola
DEC IBM, Proteon
16.
17.
18.
13. Data Switch, Proteon
14. Ascend Communications, AT&T
15. Banyan, DEC
16. Retix. Novell
17. Cheyenne Software, Networth
18. ACC Network Systems, UB
Networks
19. Stratacom, Vitalink
Communicstions
19.
Data General, Proxim
ADC Kentrox, Ericsson
Dynatech Communications. General
Datacomm
Interphase, Network Equipment
Technoloies
Data Switch, Lannet
Axon Networks, Chipcom
Gandalf Technologies, Microcom
Artel Communications, Optical Data Systems
Alcatel Telecom, Newbridge Networks
Banyan Systems, NCR
Fore Systems, General Instrument
BBN. Intel
Lanoptics, Cisco
Network Communications. Novell
Newport Systems, Networth
Network Peripherals, SynOptics
Alantec, Wellfleet Communications
Applied Network Technologies, Crosscom
Aim Technology. Network General
4.
5.
1.
2.
3.
AM Communications, Scientific Atlanta
Centillion Networks, Olicom USA
4.
6.
7.
Ascend Communications, PictureTel Corp.
Dayna Communications, Xircom
Cabletron, Concord Communications
NetEdge, NorTel
9.
10.
II.
12.
Light Stream, Tellabs
ShivaSpider
Ancor Communications, Sun Microsystems
Aironet Wireless Communications, Digital
5.
8.
CentillonNetworks,
FirstVirtualorp
Ascom Timepe, Netri, Pacific
Communication Sciences
Ocean
15.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
Note: Firms highlighted in bold have centrality above the means and those underlined have centrality one
standard deviation above the means. The subgroups in each period were identified directly from Figures 5,
6, and 7.
There are 40 and 21 alliance agreements in the core and the periphery structures
respectively but the two experts were not aware of this dichotomy. Each of the two
32
A summary of other alliance agreements is included in Tables E- 1, E-2, and E-3 of Appendix E.
36
expert coders spent approximately three hours in coding the agreements, separated in
time and place. Among 61 alliance agreements, one was discarded from the sample since
both coders identified that it was unrelated to any form of technological development. El
(the engineer) was unable to determine the nature of 21 agreements but he attempted to
code 8 of them anyway. However, E2 (the research scientist) coded all 60 agreements.
Among 47 dual coded agreements, 17 (36%) were coded as modular by one coder and
incremental by the other. El also identified four architectural innovations which E2
coded as either incremental or modular innovations. Columns A and B of Tables D-1, D2 and D-3 show the coding by E2 and El respectively.
In fact, after coding both experts raised some concerns about their judgments,
which might explain the variation in coding. They were not sure whether to assess the
technical outcomes of joint development activities or the original technologies brought by
individual partners. Joint development activities were critical to the original innovations
when the concepts of these innovations were still relatively new. However, the same
activities were later considered trivial after such innovations had been widely spread.
Based on his in-depth knowledge of the industry history, E2 concluded that the timing of
joint development activities relative to the development of the original technologies is an
important criterion.
He made his decisions using the following guideline: if the time
interval between the original and the joint development is within a couple of years, the
characteristics of both activities would be considered together in coding the agreement;
otherwise, the joint development activities alone would be coded within the taxonomy of
37
innovation in Figure 3.
Another concern raised was whether to consider either or both the technical and
market implications of innovations.
However, such criteria were not obvious in the
taxonomy of innovations. Both experts therefore coded the agreements strictly based on
the technical changes implied by the relevant activities.
To reduce the errors in coding, E2 offered to reassess all 21 agreements found
disparate in both sets of answers.
Prior to the second exercise, we also identified 10
additional agreements which created inconsistency in E2's coding.
E2 finally went
through 31 agreements the second time and found that he needed to change 5 (16%) of
his original answers. Two of these changes are now consistent with the results by El but
three others are not. Unfortunately, El was not available to do the same exercise and so
further comparison is not possible. Given that the results by El are incomplete, Table 8
shows the statistical summary of the results by E2 only.
Table 8. The Nature of Innovations Associated with the Core-Periphery Structure
Table D-1 (1985-1989)
Table D-2 (1990-1993)
Table D-3 (1991-1996)
Arch.
Mod.
Incr.
Arch. Mod.
Incr.
Arch.
Mod.
Incr. Total
Core
0
4
2
0
5
6
I
12
9
39
Periphery
0
0
2
0
2
8
0
4
5
21
Total
0
4
4
0
7
14
1
16
14
60
Note: Arch. = Architectural innovations, Mod. = Modular innovation, Incr. = Incremental innovation.
These results are coded by E2.
Only one architectural innovation is identified in all agreements, achieved within
the core structure of firms.
The core structure also consists of 21 (54%) modular
innovations and 17 (44%) incremental innovations.
38
The distinction between the two
numbers appears to be marginal but the number of modular innovations has increased by
more than 100% from the second to the third period.
On the other hand, 6 (29%)
modular innovations and 15 (71%) incremental innovations are identified in the periphery
structure. The analysis of variance (ANOVA) shows that the core-periphery structure of
the industry can indeed be differentiated by modular and incremental innovations at the
5% statistical level.
In technology alliances, core firms were more likely to develop
modular innovations whereas periphery firms develop incremental innovations.
The above results raise several questions about the relationship between coreperiphery structure and modular-incremental innovation. Do firms that persist in alliance
formation also tend to develop modular innovations internally? Likewise, do firms that
seldom form alliances also tend to focus on incremental innovations? In the two experts'
opinion, an attempt to classify "modular firms" versus "incremental firms" may prove to
be futile since many technical details internal to the firms are not obvious nor available
for systematic analysis.
Perhaps a more relevant question is whether modular
innovations are better achieved through partnerships for certain firms and incremental
innovations for others?
A modular innovation replaces the core concept of a component layer with
completely new functionality but leaves other layers unaffected.
Modular innovations
thus have considerable implications on cost-performance and interoperability (or
compatibility) across heterogeneous network environments, subsequently affecting many
competitors, suppliers as well as customers (Christensen 1992a).
39
As such, forming
alliances to develop modular innovations with competitors, suppliers or customers can
gain greater support for interoperability that is demanded in the entire market.
Cooperation with potentially "powerful" partners may even enhance the visibility and
status of the focal firm in the industry (Baum & Oliver 1991).
The alliances formed between 3Com and Microsoft to develop OS/2 LAN
Manager in the late 1980s probably best illustrate the above proposition. Their modular
innovations, as shown in column A of Table D-1, were networking software compatible
with the IBM networking environment.
Robert Potter, former president and CEO of
Datapoint Corp., which commercialized the first LAN technology ARCnet, observed,
"[The biggest story of 1988 was] the rapid movement of PC network software vendors
away from proprietary protocols and software and toward the IBM and Microsoft
standards." 3 3
In contrast, incremental innovations add new functionality to a component layer
and enhance interfaces with adjacent layers.
Little disruption to the basic network
architecture will be incurred. For example, the first four agreements under the periphery
structure indicated by Column A of Table D-2 illustrate the implementation of new
functionality in established technologies such as bridges and routers (which are Layer 2
and Layer 3 technologies respectively) to support FDDI or Frame Relay (which are both
Layer 2 protocols.)
Besides the nature of the innovation, are there other factors that possibly explain
the correlation between core-periphery structure and modular-incremental innovation?
40
Assuming that specific position of the firm in the network can influence the technological
options pursued by the firm, different insights will accrue to individual firms in the same
network (Burt 1992b). Individual firms have made investments in their relations with
others and developed knowledge about their parts of the network (Hakansson and
Johanson 1988). Increasing networking experience and information benefits bring better
opportunity to implement competing and/or complementary technologies. In contrast, the
lack of such opportunity limits the extent to which internally developed technologies can
affect the market. Thus, firms in the core of the network are better positioned to jointly
develop innovations with greater technical implications than those firms in the periphery
of the network.
If, however, the relationship between core-periphery structure and modularincremental innovation is a spurious one, factors that may cause the two events
independently are the costs and complexity of existing technologies. When such costs
and complexity are high, firms tend to share risks and costs with their partners
(Contractor & Lorange 1988). Under similar conditions, more modular innovations are
undertaken to replace mainstream technologies within a component layer. Even when a
firm's innovations replace its old technologies, the firm can gain a first-mover advantage
in entering new markets (Christensen
1992a).
The continuing growth of Sun
Microsystems in rapidly changing environment can in fact be explained by the firm's
ability to introduce "economies of substitution", whereby technological development
involves replacing
33
only certain components
NetworkWorld, December 26, 1988.
41
while retaining others
(Garud &
Kumaraswamy 1993).
Under these circumstances, neither network position nor the
nature of the innovation has causal implications for one another.
Given that core-periphery structure and the nature of the innovation may covary
because one variable causes the other or because a third variable causes both, the study
does not have sufficient data to prove one or the other of these three possibilities.
However, the stability and the significance of core-periphery structure have raised many
important implications for future studies of alliances and innovations. The instability of
some firms in alliance formation may not be just a case of random noise but instead
reflects a distinction between a core set of relationships that are important in the industry
and a periphery set of relationships that are important only on some occasions.
7.
Discussion and Conclusions
With
12 years of alliances among the 150 sample firms in the data
communications industry, the longitudinal data in the study have presented strong
evidence for some regularity and stability of the industry network. The network structure
is best described in terms of core-periphery, whereby a small number of firms are seen
repeatedly in the core of the network and many others are in the periphery forming
sporadic relationships.
The core-periphery structure appears to be fluid in that some
firms enter and exit at the periphery of the network whereas others move from the
periphery and then remain in the core of the network. Instead of being the targets for
acquisition, members of the network have become more likely to acquire non-members.
42
Moreover, core firms are more likely than periphery firms to make acquisitions.
Several interesting questions have been raised about the motives for alliances and
acquisitions underlying the core-periphery structure.
In social network theory, the
emergence of networks is a result of resource mobilization in the organizational field
(Granovetter 1992). As discussed in section 5, sources of network stability are related to
one or more factors, namely, transaction costs, information benefits and degree of
resource dependency. In addition, the acquisition activity in the industry does not appear
to have an impact on network stability. The sociological argument further informs us that
the sources of network variability are contingent upon the specific position of the firm in
the social network.
More important, the innovative behavior of the firm can be
influenced and constrained by its own set of past relationships.
Further analysis in Section 6 based on limited data shows that the set of core ties
is associated especially with modular innovations whereas the set of periphery ties is
associated more frequently with incremental innovations.
Neither core nor periphery
alliance relationships have generated radically new innovations and few architectural
innovations are found. What about the nature of technologies possessed by the 57 nonmembers and 13 members of the industry network that were acquired by both core and
periphery firms? In addition to some incumbent firms in the industry, the set of nonmember firms also includes new entrants.3 4 Existing studies of technological innovation
assert that new entrants are more likely to have radical and competence-destroying
innovations (Tushman & Anderson 1986; Utterback & Suarez 1993).
43
Although such
innovations have not been detected in this research, it is conceivable that acquisitions
have been made in order to invest in such potentialities.
Since the search for more
qualified and willing experts takes time, the attempt to code the acquired technologies
will be explored in future research.
The above concern in fact points out a major limitation of the study in its original
intent to exclude analyses of firms not participating in the network driven by technology
alliances. The difficulty of conducting research built upon social network theory is to
define the boundary for a system of organizations and to link the structure of social
relations to the behavior of the firm (Marsden 1990). "Is the population the right unit of
analysis?" was raised by DiMaggio (1994), a question remaining most problematic and
yet challenging to researchers who are interested in the sociology of organizations. What
defines the boundary of the network in the study is an arbitrary artifact which could be
eliminated if all 496 industry firms are included in the sample. Those firms that have
been identified as non-members can be viewed instead as potential members of the
industry network. Moreover, about 53% (85 out of 160) of the acquisitions from 19851996 were made among the non-members. Analyses based on the population data would
no doubt enlighten us on all the questions uncovered in this study.
In spite of the
limitation, the findings presented so far remain valid for the core-periphery structure of
interorganizational networks.
Existing studies of technological innovation have identified other patterns of
network structure.
34
Barley, Freeman and Hybels (1992) found in the biotechnology
The term new entrants refers to start-up firms and established firms from other related industries.
44
community that their clustering results are consistent with the generalist-specialist
structure in population ecology.
Moreover, variation in alliance activity is associated
with firms whose attributes differ in a coherent fashion. In contrast, Kogut, Shan and
Walker (1992) showed that network structure rather than firm attributes affects the
decision to cooperate in the biotechnology industry. In their follow-up study, they found
that the structural position of the firm, as a proxy for the amount of social capital, is a
strong predictor of network formation (Walker et al. 1997).
Finally, based on the
similarity of firm capabilities, Nohria and Garcia-Pont (1991) identified complementary
blocks and pooling blocks as significant structural patterns of the automobile industry
network. The above studies have alluded to network positions and firm attributes as
potentially important factors in interorganizational dynamics.
Besides network positions and firm attributes, the highly interdependent, complex
and yet standardized properties of the data communications technologies appear to be
additional conditions for alliances and innovations. Successive innovations in this market
are defined by both how well they interoperate with and how much they differentiate
from competing products. In many instances, alliances bring not only complementary
technologies but also competitive information about competitors' products.
The
description of technological innovation here is akin to the concepts of "new technological
systems", "large technological systems" and "open systems" explained by Freeman
(1994), Hughes (1987) and Tushman and Rosenkopf (1992) respectively. Tushman and
Rosenkopf (1992:312) argue that "the more complex and/or open the product, the greater
45
the technical uncertainty and the greater the intrusion of organizational dynamics in
technological evolution."
They further posit that interorganizational dynamics are
heightened only when core technologies are threatened (Rosenkopf & Tushman 1994).
The findings in this study have implied that firms in the core of the network seek new
technological opportunities through horizontal cooperation. These findings are also in
accord with the observations made in the development of complex flight simulators
(Miller, Hobday, Leroux-Demers & Olleros 1995).
Therefore, the core-periphery
structure of the industry network may not be unique to the data communications industry.
8.
Directions for Future Research
The concept of interorganizational networks is not yet fully defined and
understood in the study. A good definition of interorganizational networks is one that
refers to its significant structural patterns, the governance mechanisms, as well as the
conditions underlying network formation, stability and decline.
Jones, Hesterly and
Borgatti (1997) broadly define networks of firms in the following: "network governance
[that] involves a select, persistent, and structured set of autonomous firms (as well as
nonprofit agencies) engaged in creating products or services based on implicit and openended contracts to adapt to environmental contingencies and to coordinate and safeguard
exchanges. These contracts are socially and not legally binding." As demonstrated in the
study, short-term formal agreements involving technological development also have
potential impact on emerging relations in the industry. Therefore, two general research
46
questions are: 1) What are the differences between formal and informal networks of
firms? 2) Under what conditions is one form of networks more salient than the other?
Jones and her colleagues argue that better understanding of these conditions have
implications on adaptation, coordination and safeguarding in network governance.
A more specific research question arising from this study is what factors
determine the role between core and periphery firms (or high-centrality versus lowcentrality) and how these factors shape the emergence of new technologies.
If
technological attributes are fundamental to the dynamics of core-periphery network, the
interaction and replication processes among organizations add a new dimension to the
evolution of new technologies (see Figure 8). Central to the co-evolutionary concept of
organization and technology are the interactions between organization and technology
and the feedback of their interactions to firm behavior and technological attributes (Baum
& Singh 1994; Rosenkopf & Tushman 1994; Van de Ven & Garud 1994).
From a
system dynamics viewpoint, both positive and negative feedbacks are possible in the
relationship between organization and technology. Understanding both technological and
organizational attributes will shed light on the conditions under which positive feedback
loops are more likely than negative feedback loops.
Therefore, the co-evolutionary
setting for an open, complex technological system may extend the linear models of
"dominant designs" and stage-evolution explained in Abernathy and Utterback (1978)
and Tushman and Anderson (1986).
47
Figure 8. Dynamic Feedback Loops of Organization and Technology
Technology
Organization
However, an attempt to draw any theoretical implications of the co-evolutionary
process is both difficult and frustrating since our understanding of the potential impact of
interorganizational relationships on technological performance is still limited.
More
empirical research using rigorous statistical methodology is required to examine the roles
of core and periphery firms in affecting their technological outcomes.
In fact, the
dynamics of core-periphery structure is consistent with the sociological explanation of
embedded relations as the social capital of the firm (Burt 1992a; Granovetter 1992). The
social capital of the firm is the amount of resources that have accrued to the firm by
virtue of maintaining direct and indirect relationships with other firms in the same
organizational field. The behavior of the firm, and hence its technological performance,
is constrained and influenced by its ability to gain access to the social capital in the
environment.
On the contrary, alternative theories suggest that firm attributes might be a more
important determinant of firm behavior. The resource-based view of the firm infers that a
firm's internal resources are distinctive capability sustaining the competitive advantage of
the firm (Barney 1991; Wernerfelt 1984). Similarly, Teece (1986) argues that a firm
48
must have control over the necessary complementary assets in order to implement its
innovation successfully. Yet in another viewpoint, Rosenbloom and Cusumano (1987)
posit that learning and consistency of management strategies are key to successful
innovation.
While attributes of firms such as internal capabilities and resources are
important sources of innovations, the differential access to information and resources
through interorganizational relationships is also a potential source of firm heterogeneity
that cannot be ignored. Therefore, an empirical contribution to the field of economic
sociology and technology management would be to assess the relative importance of firm
attributes versus structural position in affecting technological performance of the firm.
49
Appendix A. Synopsis of Local Area Network Standards and Technologies
Local area networks or LANs are used to connect computers located within a single
building or a localized group of buildings.' PC-based LANs generally include a network file
server, network operating system, cabling and connectors, adapter cards, shared peripherals
such as printers and facsimile machines, and applications software. Adapter cards that support
the type of standard protocol configured in a network, like 3Com EtherLink PC card for
Ethernet network, provide the basic connection from PCs to the network cables.
Internetworking devices such as bridges, routers, gateways, hubs, and switches are installed
when data exchange between networks is necessary. Figure A-1 below shows three LANs
being connected to a backbone LAN by two bridges and a router. The backbone in turn is
connected to a wide area network called PSDN by a switch.
The early application of LANs was to enable distributed workstations to access an
electronic mail server or a laser printer. Today, sophisticated applications involving the
transmission of documents incorporating images and other media are becoming common,
increasing the demands on the capacity of LANs. Devices known as switches are used to
interconnect LAN segments in order to meet the demand for higher bandwidth. For greater
physical separations of LAN segments, high speed LANs will be used as backbone. Another
type of network is wide area networks or WANs, which are composed of multiple
establishment-wide LANs. WANs can be classified into two categories: public data networks
(PDNs) and private networks. PDNs are installed and managed by the public carriers while
private networks are by large national and multinational corporations. Private networks are
also known as enterprise-wide networks.
The two LAN standards that have been widely adopted are Ethernet-based (such as
10Base-T) and Token-ring networks. As for internetworking between LANs and WANs, one
of the important considerations is the standard for providing end-to-end network services to
local users. End-to-end network services are delivered such that the possible presence of
multiple networks is made transparent to the users.
At present, two competing
internetworking standards provide these network services, namely, ISO Internet Protocol, 2 and
A useful academic reference text is by Larry Peterson and Bruce Davie, "Computer Networks," 1996,
San Francisco, California: Morgan Kaufmann Publishers, Inc.
2 Founded in 1946, ISO is an international standards organization composed of national standards
50
Internet Protocol (IP) that is used by the Internet (formerly known as ARPANET). Issues that
are related to how network services should be provided are called network management,
which includes network maintenance, addressing, routing, error and flow control.
Owing to the complexity of data communications, a reference model for the structure
of a communication system has been formulated so that the standards associated with different
levels of the system can be simplified. Figure A-2 below presents the International Standard
Organization (ISO) Reference Model for data communications between two computers.
Figure A-3 illustrates the network environment of the ISO Reference Model and the
technologies which are included for this study.
The ultimate goal of interconnectivity is to achieve an open system interconnection
environment. That is, a set of application programs, running on different computers connected
by proprietary networks, are enabled to perform distributed applications. This high level of
connectivity
is further supported by
two different
standards,
i.e.,
Open
Systems
Interconnection (OSI) suite and Transport Control Protocol/Internet Protocol (TCP/IP). Both
standards support specific application protocols such as remote access and mail service. The
most popular applications currently supported by TCP/IP, for example, are electronic mail
system and Netscape.
bodies from over 75 countries. For example, ANSI (American National Standards Institute) is a
member of ISO. ISO has defined a number of important computer standards, the most significant of
which is perhaps OSI (Open Systems Interconnection), a standardized architecture for designing
networks. See http://www.sandybay.com/pc-web/ISO.htm.
51
Figure A-i. LAN-based Data Communications Networks*
Figure A-2. International Standards Organization (ISO) Reference Model*
Computer A
Computer B
--
I
Application Layer
Application Layer
Presentation Layer
Presentation Layer
Session Layer
Transport Layer
Session Layer
Transport Layer
Network Layer
Network Layer
Link Layer
Link Layer
Physical Layer
Physical Layer
r
I
Data Network
Network environment
OSI environment
Real systems environment
52
1
0
Figure A-3. Network Environment - Standards and Technologies*
4 Transport Layer
Transmission Control Protocol (TCP)*
or
User Datagram Protocol (UDP)
3 Network Layer
Internet Protocol (IP)
2 Data Link Layer
Logical Link Control
IEEE 802.2
Medium
Access Control
IEEE 802.3
Ethernet
IEEE 802.4
Token Bus
Transmission Medium
*
Bridges. Switches function
with dissimilar LANs
IEEE 802.5
Token Ring
I
1 Physical Layer
Gateways works at
Network Layer and
above
Bridges, Switches
function with similar
LANs;
Routers work in both
Network and Data Link
Layers
Repeaters and network interface
cards operate at the physical layer;
Coaxial cable is by far the most
popular medium implemented for
LANs
All three figures are adapted from Fred Halsall, "Data Communications, Computer Networks and
Open Systems," 1996, (Fourth Edition), Addison-Wesley.
53
Appendix B. Social Network Analysis using UCINET IV3
Output of Geodesic Distances among Pairs of Firms, 1985-1988
Table B-1.
DISTANCE
ADJACENCY
Type of data:
NONE
Nearness transform:
C:\UCINET\NETWOR KI\D85-8LSC
Input dataset:
For each pair of nodes, the algorithm finds the # of edges in the shortest path between them.
Length of optimal paths between all pairs of nodes:
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
5
6
7
3
4
1
2
3COATTCHPCODCPQCDSDCADECDSCEXC HPIBMINTMDGMCT MSNETNVLORCRABRTXSCOSTRSNPSYTUBNWSDSMC
13C0
2 ATT
3 CHP
4 COD
5 CPQ
6 CDS
7 DCA
8 DEC
9 DSC
10 EXC
11
HP
12IBM
13INT
14MDG
15 MCT
MS
16
17NET
18NVL
19 ORC
20RAB
21RTX
22SCO
23 STR
24SNP
25SYT
26UBN
27 WSD
28SMC
0
3
4
1
4
4
1
3
4
3
1
3
4
3
4
2
4
5
4
5
3
3
4
6
2
4
7
5
4
3
0 7
7
0
4 3
7
2
7
2
4 5
6 1
7
2
2
7
5
2
7
2
1 8
7
2
1 8
1 6
3 8
9
4
8
3
4 9
7
2
2
7
7
2
5 10
5 2
3 8
611
2
9
1
4 4
4
7 7
2 2
3
3 3
0
3
0 2
3
2 0
2
5 5
2
1 1
2 2
3
7 7
4
2
5 5
7 7
4
5 8 8
7
4
7
5 8 8
3
6 6
5 8 8
6 9 9
5 8 8
6 9 9
7
4 7
4
7 7
3
2 2
7 10 10
1 2 2
5 8 8
8 1111
6 9 9
4
1 3
4 6
7
5 1
2
3
2 2
2
5 1
2
5 1
5
0 4
1
4 0
0
5 1
7
4 6
5
2 4
6
7
4
5
7 8
7
4 6
7 8
5
6
3 5
7 8
5
6 8 9
7 8
5
6 8 9
7
4 6
7
4 6
2
5 1
7 9 10
3 1 2
7 8
5
81011
6 8 9
3
2
7
4
7
7
4
6
7
0
2
2
3
2
3
1
3
2
1
4
2
2
7
3
5
3
4
4
3
2
7
4
7
7
4
6
7
1
2
5
2
5
5
2
4
5
2
2
0
2
3
2
3
1
3
4
3
4
2
2
5
5
3
3
6
4
2
0
3
2
3
1
1
4
3
4
2
2
7
5
5
3
6
4
4
1
8
5
8
8
5
7
8
2
3
3
2
3
3
0
3
2
2
4
5
4
5
3
3
8
6
6
4
7
1
4
1
8
5
8
8
5
7
8
3
2
7
4
7
7
4
6
7
1
3
4
3
4
2
1
3
1
3
4
3
4
8
6
8
6
3
9
1
76
6 4 6
5
3 4
2 4 1 3
5
7 2
6 7
5 6 5
4 3 3
7
5
6
4
2
4
2
0
4
3
2
7
4
7
7
4
6
7
2
2
3
2
3
3
5 3
2 4
4 3
4
0
3 4
2
4
8
9
3
3
75
1
3
4
4
6 4
5
4 7
4
6 2
3
6 2
7
5 5
9
810
2
811
2
6
5 8
3
7 1
810
2
811 9
2
811 9
2
810
2
6
5 8
5
7 3
710 8
1
9 1
810
2
811 9
2
3 44
7 3 5
3
64
5 3
2 5
64
7 5 5 3
471
8 6 6
6 4
7 5 5 3
4
7 3
8 6 6
2
5 3
4 4
1 6
8 6
6475
7 1 26
9
1
4
6 3
8 3
6
7 1 4
9
3
5 3
6 4
5
2 7
4
3 6
7 5 5
811
9
7 010 2
17
8 2
510 0
7
0 6 9
2 8
5
3
8
2 6 0
9 3
0 8
611 1
0
9
7 7 5 8
3
2
7
4
7
7
4
6
7
2
2
4
4
5
4
5
3
4
3
4
5
0
2
2
3
4
3
3
2
0
5
4
5
2
3
2
3
4
4
5
4
5
5
4
9
6
9
9
6
8
9
4
3
8
5
8
8
5
7
8
1
2
1
4
3
4
0
2
5
4
9
6
9
9
6
8
9
3
1
1
2
1
2
4
5
4
5
2
3
1
3
3
2
3
0
2
3
0
3
4
3
8
5
8
8
5
7
8
2
1
6
3
6
6
3
5
6
2
3
2
3
1
3
4
3
4
0
0
7
5
5
3
6
4
Distance matrix saved as dataset C:\UCINETNETWORK9\D85-8D
Elapsed time: I second. 1/29/1998 9:37 AM.
UCINET IV 1.66/X Copyright 1991-1996 by Analytic Technologies.
3 Borgatti, Everett and Freeman, 1992, UCINET IV, Columbia: Analytic Technologies.
a software program for social network analysis. For more information see
http://www.heinz.cmu.edu/projectINSNA/softinf.html
54
UCINET IV is
Table B-2.
Output of Multidimensional Scaling (MDS) Coordinates among
Pairs of Firms based on Geodesic Distance, 1985-1988
NON-METRIC MULTIDIMENSIONAL SCALING
Input dataset:
C:\UCINET\NETWORK9\D85-8D
TORSCA
Dissimilarities
Starting config:
Type of Data:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
3CO
ATT
3C1
COD
CPQ
CDS
DCA
DEC
DSC
EXC
HP
IBM
INT
MDG
MCT
MS
NET
NVL
ORC
RAB
RTX
SCO
STR
SNP
GEl
TND
WSD
SMC
1
2
-0.30
0.32
-1.47
-0.60
-1.49
-1.45
-0.40
-1.18
-1.36
0.48
0.00
0.50
0.30
0.46
0.21
0.30
0.79
1.00
0.56
1.20
0.27
0.31
-1.41
1.21
-0.89
0.90
1.37
0.36
-0.02
0.51
-0.21
-0.07
-0.09
-0.31
-0.09
-0.24
-0.54
-0.05
0.06
0.35
0.84
0.18
0.78
0.17
0.56
-0.46
-0.34
-0.20
0.29
0.10
-0.42
-0.70
-0.16
-0.18
-0.94
1.18
Coordinates saved as dataset C:\UCINET\NETWORK9\D8COORD
Note: Several starting configuration methods to locate the firms are available under non-metric MDS
technique in UCINET: Metric -Gower's metric coordination procedure; Torsca - principal components
of rank-order data; Random - locates points randomly in space. Torsca was applied to the dataset in
this study. See Table B-3 for MDS output.
55
Table B-3.
Output of MDS Map based on Coordinates in Two-dimensional
Space
Dim 2
1.26
SMC
0.66
MCTNT
ATT
RTX
0.06
NET
IBM
MDG
SCO
HP
CPQ
3C1
COD DCACO
EXC
GEl
TND
RAB
DEC
-0.54
CDS
STR
DSC
ORC
NVL
SNP
WSD
-1.14
-1.13
-0.55
0.03
0.61
1.19
Dim 1
Stress in 2 dimensions is 0.041
Elapsed time:
1 second. 1/29/1998 8:49 AM.
UCINET IV 1.66/X Copyright 1991-1996 by Analytic Technologies.
56
Appendix C. Centrality
In social network analysis one of the commonly used measure is centrality, which can
identify the most "important" or "prominent" actors in a social network.
According to a
review of social network analysis (Freeman 1978), the idea of centrality as applied to human
communications was introduced by Bavelas at the Group Networks Laboratory at M.I.T. in
the late 1940s. The studies conducted by the research group under the direction of Bavelas
found that centrality was related to group efficiency in problem solving, perception of
leadership and the personal satisfaction of participants (Bavelas 1948; Bavelas & Barrett
1951).
Three types of centrality measures have been used to quantify the prominence of an
individual actor embedded in a network: degree, betweenness and closeness (Freeman 1978;
Wasserman & Faust 1994).
Degree centrality of an actor is the simplest measure of the
number of ties (degree) to all other actors (nodes) in a graph.
The most central firm is
considered to be the most active actor in the network. Betweenness centrality measures the
number of intermediaries between two nonadjacent actors. These "other intermediate actors"
potentially might have some control over the interactions between the two nonadjacent actors.
If an actor lies between many of the actors via their shortest paths, it is said to have high
centrality in the network.
Closeness centrality measures how central or close an actor is to all other actors. This
is the same as finding minimum distances to all other actors and the idea is that the more
central the firm the lower the cost or time for communicating information to and from all other
actors. Assuming that the shortest (geodesic) distance between two actors is d(ni, n), the sum
of the geodesic distances from actor zi to all other actors lnjis given by
g
d(ni, nj)
where j
i.
ji=
For example, in a star network with 5 actors connected to a single actor in the center,
the sum of the geodesic distances for the central actor is 5 (1 + 1+ +1+ 1) whereas the sum for
each of the 5 other actors is 9 (1+2+2+2+2). The closeness centrality index for actor ni based
57
on Sabidussi's method (Freeman 1978) is given by
C(ni) =
d(ni, ni'
The closeness centrality index generated in UCINET IV is normalized by dividing the
actor's index by the maximum possible closeness centrality index, i.e. (g-l)-', in the network.
The standardized index ranges between 0 and 1. When the index equals unity, it means the
actor is maximally close to all other actors. When the index equals zero, one or more actors
are not reachable from the actor in question. However, the closeness centrality measure is
only meaningful for a connected graph, i.e. every node is reachable from all other nodes.
Otherwise, if one node is not reachable, then the distance from all other nodes to this specific
node is infinitely long and the distance sum for every actor is oo. The input to closeness
centrality is a symmetric binary matrix, i.e. when the number of alliances is 1 or greater
between two actors, the value will be 1, otherwise 0.
58
Appendix D. Cohesive Subgroups and Technological Developments, 1985-1988,
1989-1992, 1993-1996
Table D-1.
Cohesive Subgroups and Technological Developments, 1985-1988
Cohesive Subgroup
Technological Development
Coder
Ratings
A B
In 1985, Intel, Microsoft and UB Networks announced a strategic development and
marketing agreement for Intel's Open Net system, one of the key features of which is
compatibility with PC Network by IBM. In 1988, UB Networks signed an OEM
licensing agreement for Microsoft's LAN Manager and intended to extend its current
product capabilities to an OS/2 base.
In 1985, Novel and 3Com formed agreements with Microsoft to obtain MS-DOS 3.1
and MS Network software, designed to be closely compatible with IBM's PC-DOS 3.1
and PC Network software. In 1987, 3Com jointly developed the OS/2 LAN Manager
with Microsoft; in addition, both companies jointly made sales calls to large corporate
users. 3Com also developed 3 open, network system software based on the LAN
Manager.
NET has benefited from its 2-year-old joint development and marketing relationship
with IBM in the T- I market.
In separate cooperation with Sytek and 3Com, Codenoll has come up with fiber optic
versions of the two companies' network adapter cards.
AT&T and Intel jointly developed their integrated services digital network (ISDN)
microchips compatible in a move that would give ISDN terminal makers more choices
in designing products.
Novell's agreement with Oracle used Excelan's TCP/IP Workstation software and
Novell's TCP/IP Gateway software to provide transaction processing through links
between LANServer and PC workstations. Oracle customers will be able to use Netware
to connect with a variety of network schemes, including Ethernet, token-ring networks
and Arcnet.
M
I
M
I
I
?
M
M
M
M
I
M
I
M
I
M
CORE
1985, 1988 Microsoft
and UB Networks
1985, 1987 Microsoft
and 3Com
1988, IBM and Network
Equipment Technologies
1986, Codenoll
Technology and Sytek
1986, AT&T and Intel
1987, Oracle, Novell,
Excelan
PERIPHERY
1988, SynOptics and
Western Digital
1988, Chipcom and DEC
SynOptics' LattisNet 10-Mbps Ethernet transceiver will be incorporated into a version
of Western Digital's EtherCard Plus network interface card for the IBM PC XT and AT,
PS/2 Model 30 and compatibles.
The Ethermodem III Bridge is the result of a joint development agreement between
Chipcom and DEC. Based on DEC's LAN Bridge 100 architecture and Chipcomr's
broadband transceiver technology, the bridge links Ethernet subnetworks to a broadband
Ethernet backbone.
Source: Lexis/Nexis, 1996/7.
Note: A is E2 (Research Scientist) and B is El (engineer). M - Modular Innovations, I - Incremental
Innovations, A - Architectural Innovations
59
Table D-2.
Cohesive Subgroups and Technological Developments, 1989-1992
Cohesive Subgroup
Technological Development
Coder
Ratings
A
B
The three companies put together LAN II, consisting of multiple protocols and multiple
networks. It's a workstation solution intemetworked with a mainframe solution. NSC
provided the HyperChannel series of high-speed networks used to connect
supercomputers and mainframes. Excelan is focused on networking workstations,
minicomputers and PCs. Wellfleet developed Ethernet bridges and routers for linking
M
M
M
M
M
M
I
I
I
M
I
I
I
M
M
M
M
M
l
I
I
?
CORE
1989, Excelan, Network
System Corp. and
Wellfleet
Communications.
LANs and WANs.
1990, Cisco and Madge
HP and 3Com developed products together and cooperated in worldwide joint service.
In promoting single interface, HP embraced 3Com's networking scheme to tie its
minicomputers to other vendors' hardware. The firms announced six areas of product
development focused heavily on standards such as TCP/IP, CCITTs X.400 and the
Open Systems Interconnect model.
DEC offered a licensing program for its Local Area Transport (LAT) protocol, saving
third-party vendors the trouble of having to reverse-engineer the popular terminal-toVAX networking specification. Banyan integrated networking products with LAT.
Cisco licensed Madge's Fastmac data buffering and transfer software for use in Cisco's
Networks
Token-Rine routers.
1990, 1991, Cisco and
SynOptics
In 1990, they jointly developed a router module for Synoptics' LattisNet wiring
concentrator. In 1991, SynOptics, Cisco and SunConnect jointly developed the nextgeneration, integrated routing hub for Ethernet. Token Ring, and FDDI, dubbed the
RubSystem. In another joint development agreement, Cisco and Synoptics integrated
routers with the wiring hub.
Madge announced a licensing agreement with SynOptics for Madge's FastMAC TokenRing software. FastMAC Token-Ring and Bridge Management software package
would be incorporated into SynOptics' LattisNet Token-Ring product line.
IBM and Chipcom jointly developed intelligent hubs to link heterogeneous computers
and networks into enterprise networks. The partnership helped IBM fill a gap in its
internetworking product line.
NS agreed to license ATM interface designs from Adaptive Corp., a unit of NET, and
used the technology to develop multimedia computer and high-speed networking
1989, 3COM and HP
1989, Banyan and DEC
1991, Madge Networks
and SynOptics
1992, Chipcom and IBM
1992, National
Semiconductor and NET
equipment.
1990, 1992 Retix and
Novell
1991, Tiara and Sun
(Sitka)
1992, Ascend Comms.
and AT&T Paradyne
In 1990, Novell and Retix in joint development to provide X.400 pathway in MHS
LAN. In 1992, Retix joined with Novell to deliver the first X.400 capability to the
NetWare Global Messaging environment.
Sitka Corp is formerly the TOPS Division of Sun Microsystems. Sitka and Tiara
improved the interoperability of two popular network operating systems and provided
each system with connections to a wider variety of network environments.
AT&T Paradyne and Ascend cooperated in the development and distribution of
products for wide-ranging networking applications in the global bandwidth-on-demand
marketplace.
60
Table D-2.
Cohesive Subgroups and Technological Developments, 1989-1992,
Continued
Cohesive Subgroup
Technological Development
Coder
Ratings
A
B
UB Networks to integrate Advanced Computer Communications (ACC) multiprotocol
bridge/router with UB Access/One intelligent hub. The agreement with ACC was
sparked in part by the need to bring branch offices into the corporate network.
Data Switch integrated Proteon's LAN products into its data processing and
communications switching and control systems.
Vitalink Communications and Stratacom had a joint-development effort through which
Vitalink bridges and routers will support Stratacom's Frame Relay Interface.
Lanex and Codenoll Technology entered into a strategic alliance to develop additional
technology and products for Fiber Distributed Data Interface (FDDI) networks.
They jointly developed GatorMIM CS media interface module. The device is an Apple
Talk-to-Ethernet media interface module, based on Cayman's gateway technology, that
plugs in to a Cabletron Multi Media Access Center Ethernet hub.
Bytex incorporated CrossCom's next generation token-ring bridging technology in its
Maestro Intelligent Switching Hub. Maestro is a matrix-switching hub that lets users
remotely configure and manage the physical links of their token-ring LANs.
The two companies jointly developed network management software for NetWorth's
10Base-T concentrators.
Netrix and Pacific Communication Sciences had a joint-development agreement that let
each company rely on the other's expertise for integrated circuit-packet switches and
voice-compression techniques.
Ericsson expanded its relationship with Luxcom to jointly develop a collapsed backbone
that supports multimedia applications.
I
M
I
M
I
M
I
I
M
M
I
M
I
I
I
A
I
A
M
M
PERIPHERY
1990, ACC Networks
and UB Networks
1990, Data Switch and
Proteon
1990, Stratacom and
Vitalink Comms.
1990 Codenoll
Technology and Lanex
1991, Cabletron and
Cayman Systems
1991, Crosscom and
Bytex
1991, Networth and
Cheyenne Software
1991, Netrix and Pacific
Communication Sciences
1992, Ericsson and
Luxcom
1992, Tellabs and Fujitsu
Network Systems
Fujitsu Network Systems and Tellabs used WAVE to make history, providing the first
mid-span meet of synchronous optical network (SONET) equipment that included a link
of data communications channels.
Source: Lexis/Nexis, 1996/7
61
Table D-3.
Cohesive Subgroups and Technological Developments, 1993-1996
Cohesive Subgroup
Coder
Ratings
Technological Development
~~______ ~~~A
B
CORE
1993, Interphase and
NET
1993, SynOptics and
Network Peripherals
1993, Networth and
Newport Systems
Solutions (NSS)
1993, Network
Communications Corp.
(NCC) and Novell
1994, Chipcom and Axon
Networks
1994, Ascend
Communications and
PictureTel
1994, Alantec and
Wellfleet
Communications
1994, Network General
Corp and AIM
Technology
1994, Crosscom and
Applied Network
Technology (ANT)
1994, Lanoptics and
Cisco
1994, LightStream
(LITES) and Tellabs
1994, 1995, Proteon,
IBM and Dec
1995, 3Com and Bay
Networks
1995, AT&T and HP
1995, 1996, Cascade
Communications and
Motorola
NET and Interphase developed ATM adapter cards to link workstations to ATM networks.
M
M
Network Peripherals and SynOptics jointly developed ATM adapter cards. Network
Peripherals built the adapters for EISA, Micro Channel and S-bus architectures using ATM
technology from SynOptics.
NetWorth licensed NSS' LAN/MPR bridge software for NetWorth's Series 4000 Local
Ethernet Bridge. Also, NSS certified the NetWorth NetWare Application Engine (NAE)
as a platform for its routers.
Novell had a licensing and technology agreement with NCC whereby NCC will acquire
exclusive rights to develop, market and support the hardware- based version of Novell's
LANalyzer network analyzer. The agreement includes the hardware-based LANalyzer
only; Novell continued to develop and market its software-only LANalyzer for Windows
product.
Joint development of two network monitoring applications for Chipcom's line of
intelligent switching hubs that complies with the Simple Network Management Protocol's
Remote Network Monitoring Management Information Base (RMON).
They have been supplying complementary solutions to their worldwide customers for
several years. The Ascend Multiband MAX Bandwidth-on-Demand product provided
public network connections for the PictureTel M-8000(2) Multipoint Conferencing family.
They had a co-operative marketing and technical support alliance, covering their Backbone
Concentrator Node routers and PowerHub products respectively. It covered four key
initiatives: interoperabilitv testing, technical training, co-ordinated technical support, and
joint sales and marketing.
They integrated their product lines to create end-to-end client/server network management.
Network General also bought a 10% stake in AIM Technology, a Unix systems
management software provider. AIM used the equity investment to accelerate
development of its SharpShooter product, a fault and performance management
application for Unix systems.
CrossComm signed a technology license agreement with Applied Network Technology for
the development of new Ethernet switching technology for CrossComm's XL product line.
M
M
I
M
I
?
I
?
M
?
I
?
I
?
M
M
I
M
M
A
M
M
9?
?
M
I
M
M
They jointly developed integrated hub/router products. The Cisco AccessPro router card
will be available with StackNet, LanOptics' multi-protocol stackable hub, and packaged
with BranchCard. LanOptics' PC based hub card.
They jointly developed and distributed ATM switching systems. The two companies
negotiated an OEM agreement whereby Tellabs would market LightStream 2010 ATM
products. In addition, Tellabs and LightStream planed a technology licensing agreement
that would allow Tellabs to develop new features and functions for the LightStrcam 2010
ATM products.
In 1994, Proteon had licensed DEC and IBM to use its internetworking software and had
extended a previous joint development agreement with IBM to develop remote site router
family products. In September 1995, IBM and Proteon reaffirmed their continuing joint
development of remote site internetworking products including implementation of the
DLSw specification.
3Com and Bay Networks jointly announced a cooperative mutual patent license
agreement. The agreement covers all patents presently held by both companies and extends
to all patents that will be issued to both organizations during the next five years.
AT&T and HP agreed to develop public and private networks using wired and wireless
technology to improve delivery of multimedia information and interactive services.
In 1995, Motorola teamed up with frame relay switch maker Cascade to develop a terminal
adapter that lets remote users run frame relay traffic over ISDN lines. In 1996, Motorola's
Multimedia and Information Systems Groups and Cascade agreed to cooperate to provide
high-speed Internet access via cable to end-users. The combination of Motorola's
CableComm system and Cascade's multiservice switching products will allow increased
capacity for high-volume, easilv expandable Internet access.
62
Table D-3.
Cohesive Subgroups and Technological Developments, 1993-1996,
Continued
Cohesive Subgroup
Technological Development
Coder
Ratings
1996, ADC Kentrox and
Ericsson
1996, NetEdge and
Nortel
A
B
M
M
M
M
I
?
A
A
M
I
M
M
Microcom formed technology partnership with Gandalf Technologies that allows Gandalf
to incorporate Microcom's modem technology into its Xpressway r, central site
concentrators.
AM Communications and Scientific-Atlanta jointly developed new transponder
technology.
I
M
I
?
They co-developed a WaveLAN driver for Banyan's VINES network operating system.
NCR also initiated a joint -marketing program with Banyan through which Banyan
M
I
M
M
M
M
I
I
l
?
Ericsson and ADC Kentrox signed a systems integration agreement to deliver ATM
solutions to network operators and service providers. Ericsson will market and sell ADC
Kentfox ATM access products for data communications and WANs. The agreement also
includes the framework for cooperative development of new ATM products and services.
Nortel's LAN Module allows network providers to offer LAN interconnect service on an
ATM-over-SONET backbone to multiple customers via the same network connection
that also supports voice and narrowband data. The LAN Module incorporates the
NetEdge's award-winning ATM edge routing technology.
1996, General Datacom
and Dynatech Comms.
They successfully demonstrated the interoperability between Dynatech's DynaStar 100
and the GDC APEX(. The combination of these products provides ATM all the way to
the access point without multiple network conversions.
1996, Fore Systems and
General Instrument
1996, BBN and Intel
They jointly developed a high-speed, ATM-based telecommunications network system.
The system is designed to enable broadband telecommunications operators to provide
high-speed, two-way communication services to computers over hybrid fiber coax (HFC)
cable TV plants.
BBN has built a trial network to demonstrate that the Resource Reservation Protocol
(RSVP) can give the guarantees necessary to ensure that real-time data, as well video,
audio and other multimedia elements, get through. The network was designed in
conjunction with Cisco, content provider Worldwide Broadcasting Network (WBN) and
Intel.
1996, Alcatel and
Newbridge Networks
They had an agreement that would enable carriers and corporate customers to deliver
SONET (Synchronous Optical Network) services from the public backbone network
directly to customer premises.
1996, Microcom and
Gandalf Technologies
1996, AM Comms. And
Scientific Atlanta
PERIPHERY
1993, Banyan and NCR
VARs can market NCR's WaveLAN wireless LAN product with VINES.
1993, ODS and Artel
Communications
1993, Data General and
Proximn
1993, Ascom Timplex,
Netrix and Pacific
Communication Sciences
Optical Data Systems and Artel integrated Artel's StarBridge Turbo Ethernet Switch
technology into the ODS Infinity Hub.
Proxim licensed its RangeLan2 kit to Data General. Data General integrated Proxim's
RaneLAN2 wireless LAN technology into a future family of mobile computing systems
Since 1991, Netrix and Pacific Communication Sciences have signed a joint-development
agreement that lets each company rely on the other's expertise for integrated circuitpacket switches and voice-compression techniques. PCSI is working with Ascom
Timeplex to develop a daughterboard that supports the CELP algorithm as well as
facsimile transmissions.
1994, Dayna Comms.
and Xircom
1994, Data Switch and
Lannet
Xircom and Dayna Communications launched a joint-development and cross-licensing
deal for Apple connectivity products
They established ajoint sales, marketing, and technology-sharing agreement to develop
and market systems to simplify and reduce the cost of integrating mainframe data centers
I
with PC-based networks.
1995, Ancor and Sun
1995, Centillion
Networks and Olicom
1996, Cabletron and
Concord Comms.
Ancor Communications and Sun Microsystems jointly developed switched Fibre Channel
support for server storage.
They developed token-ring and ATM switching solutions incorporating Centillion's
Speed Switch 100 and Olicom adaptors.
They integrated Concord's Network Health performance analysis and reporting
applications into Cabletron's Spectrum, the No. 3 enterprise management platform after
Hewlett-Packard's OpenView and Sun Microsystems' SunNet Manager.
Source: Lexis/Nexis, 1996/7.
63
M
M
I
A
I
Appendix E. Other Cohesive Subgroups
Table E-1.
Cohesive Subgroups and Technological Developments, 1985-1988
Cohesive Subgroup
1986 DEC and CDS
1988 DEC and CPQ
1988 DEC and DSC
1988 DEC and STR
1987 3CO and DCA
1988 TND and RAB
1986 MS and SCO
1988 MS and HP
1988
1988
1988
1988
MS and MDG
MS and RTX
INT and SMC
ATT and MCT
Technological Development
Concord Data Systems, Inc. has announced a multiyear agreement with DEC to develop General
Motors Corp.'s Manufacturing Automation Protocol interfaces for DEC's Microvax computers.
Compaq and DEC are developing VAX-to-MS-DOS linkages.
DEC and DSC to develop network systems; agreement designed to foster CCS, ISDN, other
telecom advancements.
DEC's EMA is OSI-based, and according to the company, will accommodate both voice and data
requirements and manage SNA and Transmission control Protocol/lnternet Protocol environments.
Stratacom will develop interfaces to EMA with its own network management products.
3Com formed alliances with Digital Communications Associates Inc. (DCA). DCA manufactures
add-on boards to link microcomputers to mainframes.
UB Networks licensed from Rabbit Software SNA technology.
Xenix-Net, developed jointly by Microsoft and The Santa Cruz Operation. Xenix-Net provides
networking, distributed file system capabilities and transparent file sharing in mixed environments
consisting of Xenix System V.2-and DOS-based systems.
HP and Microsoft announced their joint strategy for moving the OS/2 Presentation Manager to the
Unix environment.
Madge Networks licensed Microsoft's OS/2 LAN Manager.
Retix licensed Microsoft's OS/2 LAN Manager.
Intel and Standard Microsystems Corp. jointly developed LAN Ethernet VLSI.
AT&T continued to buttress its support of dominant communications protocols for its 3B processor
series through a recent alliance with Micom-Interlan, Inc. that extends Transmission Control
Protocol/Internet Protocol, or TCP/IP, functionality to the entire line.
Source: Lexis/Nexis, 1996/7.
64
Table E-2.
Cohesive Subgroups and Technological Developments, 1989-1992
Cohesive Subgroup
Technological Development
1989 CBT and LSI
Cabletron and LSI to develop a transceiver chip that complies with the proposed IEEE 10BaseT
standard for 10M bit/sec Ethernet over unshielded twisted-pair wiring.
Cabletron and Silicon Graphics developed network management system.
Cabletron and Fore Systems jointly developed products for ATM technology.
Cabletron and Windata developed a wireless LAN module for Cabletron's hub line that will forge
wireless links to workstations and other devices.
Bytex and Gandalf Data Systems jointly developed token ring solutions.
Crosscom and TRW formed alliance in Ethernet, broadband and wan internetworking solutions.
General Datacomm, Crosscom jointly developed multimedia intemetworking solutions.
Network Systems Corp and Storage Technology Corp developed storage devices for network.
Northern Telecom and Network Systems Corp. in Lan/Wan joint product development.
3Com will incorporate Synernetics Lanplex 5000 architecture into its new Linkbuilder 3GH thirdgeneration structured wiring hub.
EO licensed Sitka's (un Sun) mobile net solutions.
Netrix and Siemens Private Communications Systems jointly developed packet switching products.
HP and AMD in 10Base-T Ethernet Lan product development.
Codenoll Technology licensed from Western Digital Lan card design and software
SynOptics &Farallon Computing integrated Fs phoneNet into S's network platform
SynOptics licensed network management software from Remedy Corp
Cisco integrated its router with Ultra Network Technology's network
To develop CSCO's router to manage Cascade's frame relay network
Proteon and Netlink to develop IBM connectivity products
AT&T and Intel have signed an agreement to develop personal computers designed to support
AT&T's networked computing offerings.
DEC and David Systems joint develop SNMP (simple network management protocol) to manage
networking products
DEC and Microsoft integrated their networking strategies by linking application programming
interfaces (APIs) for DEC's Network Application Support (NAS) and Microsoft's
Windows Open System Architecture (WOSA).
Novell and Network Communications Corp developed internet test and diagnostic instruments
Novell and Microdyne jointly developed interface cards
Network Peripherals &UB Networks in FDDI concentrator products
UB Networks and BBN has been working for UB's ATM and multimedia solutions.
Vitalink and Xyplex will work together to develop and market what they claim will be the first
interoperable remote bridges from different vendors.
1991 CBT and SIL
1992 CBT and FOR
1992 CBT and WDT
1992 BYT and GDL
1989 XCO and TRW
1991 XCO and GEN
1992 NSC and STO
1992 NSC and NT
1991 3CO and SYN
1992 SUN and EO
1992 NTR and SIE
1989 HP and AMD
1989 COD and WSD
1990 SNP and FAR
1991 SNP and RMD
1989 CSO and ULT
1992 CSO and CSC
1992 PTE and NLK
1989 ATT and INT
1991 DEC and DVS
1992 DEC and MS
1991 NVL and
1992 NVL and
1991 TND and
1992 TND and
1990 VLK and
NCC
MDY
NEP
BBN
XPL
Source: Lexis/Nexis, 1996/7.
65
Table E-3 Cohesive Subgroups and Technological Developments, 1993-1996
Cohesive Subgroup
1995 BAY and MS
1995
1995
1993
1995
BAY and FUJ
XIR and NEC
TLA and TRW
TLA and PRO
1994 SUN and CNT
1993 ATT and PRE
1993 ATT and PNR
1995 ATT and MUL
1993 CSO and XER
1995 CSO and NOK
1995 CSO and TBI
1995 CSO and CPQ
1995 CSO and USR
1996 CSO and MDG
1996 CSO and XPO
1993 SHI and APP
1995 HP and NEV
1995 HP and NSC
1996 HP and CEL
1993 IBM and KAP
1996 IBM and XYL
1994 PRX and ZEN
1994 PRX and LXE
1996 ADK and PAR
1993 NET and NS
1995 DAS and DCM
1994 AXO and RAY
1994 AXO and PHO
1994 CHP and SYN
1993 ODS and SMC
1993 NEW and XYP
1996 NEW and SIE
1993 NCR and NAT
1993
1994
1994
1994
1993
NWT and TND
NWT and GRJ
SNP and STR
SNP and BA2
WFL and MAK
1994 XCO and MLT
1996 NEG and NSY
Technological Development
Bay and Microsoft formed an alliance that will integrate Bay Networks' Routing Services
(BayRS) into Microsoft's Windows NT Server.
Bay and Fujitsu Network Switching in joint development of ATM technologies.
Xircom licensed to NEC wireless technology.
Tellabs and TRW would work together to develop SONET and ATM technology and roducts.
Tellabs and Promptus Communications in joint develpment of enhanced inverse multiplexing
technology that will benefit the customers of both companies.
Sun licensed SNA network software from Computer Network Technology's Brixton Systems Inc.
Sun's Sparc/Solaris enterprise.
AT&T Paradyne developed with Premisys in network access technology.
AT&T Paradyne licensed remote access technology from Penril Datability.
AT&T Microelectronics licensed from Multi-Tech Systems the digital simultaneous voice and
data TM chip sets.
Cisco and Xerox signed a marketing agreement to resell bridges and routers. (This agreement
should have been omitted in the study.)
Cisco and Nokia to develop ATM technologies.
Cisco acquired TBIT's MICA technology and formedan alliance with TBIT.
Cisco and Compaq formed licensing and joint development agreement.
Cisco and US Robotics to integrated USR's modemand C's routing technology into I box.
Cisco and Madge Networks to develop hi-density ISDN switch.
Cisco licensed its VLAN Technology to Xpoint.
Shiva and Apple developed point-to-point protocol technology.
HP and Netvantage integrated 100VG-AnyLAN into both companies' switches.
HP and Network Systems manufactuered server channel solutions.
HP and Cellnet data system to develop wireless communications technologies and client/server
computing solutions.
IBM and Kaplana in Token Ring ATM integration.
IBM and Xylan in Lan switching product development.
Proxim and Zenith Data Sys in wireless computingploduct development.
Proxim and LXE Inc in wireless LAN product devement.
ADC Kentrox acquired 20% stake in Paragon Networks (wan) to integrate access products.
Network Equipment Technology licensed to National Semiconductor ATM software.
Data Switch's T-Bar integrated with Datacom Systems' technology to develop new Lan switch
products.
Axon and Xyplex ointly developed network manament sstems.
Axon and Phoenix Microsystems to develop products for the emerging field of LAN/WAN
monitoring system.
Chipcom and Sync Research integrated S's SNA technology.
Optical Data Sys and Standard Microsystem Corp. integrated its bridge and routing technologies
into O;s hub technology.
Newbridge and Xyplex in ATM development.
Newbridge and Siemens in ATM development.
Network Application Technology &NCR to integrate network management programs with NCR
StarSentry SNMP management platform.
UB Networks has 50.4% stake in Networth.
Networth licensed Grand Junction's router tech. 0OBase-T and Fast Ethernet adapter technologies.
SynOptics and Stratacom in ATM technology.
SynOptics and Xylogics integrated X's servers into S's hubs.
Wellfleet and Make Systems announced a technology sharing agreement aimed at producing a
network design tool for router-based networks.
Crosscom and Multimedia Communications Inc. siged a technology licensing agreement.
Network General acquired interests in Netsys network sstems.
Source: Lexis/Nexis, 1996/7.
66
Appendix F. Interoperability Forums and Standards-setting Alliances,
1989-1996
Table F-1.
Interonerabilitv Forums and Standards-settinp Alliances
Name of Alliance and Date
of Formation*
1. FDDI Forum 1989
Objective of Alliance
Established by Synernetics in 1989 to accelerate the Fiber Distributed Data
Interface (FDDI) Station Management (STM) standard to completion. 13
vendors also jointly developed a design specification for an interoperable STM
implementation that ensures full FDDI standard compatibility.
2. 10OBaseT Consortium
February 1990
3. Frame Relay Trial
January 1993
4. Fast Ethernet Alliance
August 1993
5. 100 VG-AnyLAN
September 1993
6. 20Mbps Full duplex
Ethernet
October 1993
7. ATM25 Alliance
September 1994
8. Token Ring Alliance
(Astral) November 1994
9. Versit: Convergence of
Communications and
Computing
November 1994
10. ISDN Forum
January 1996
10 vendors formed a consortium to ensure interoperability of 10BaseT products
and to exchange technical information.
13 companies participated in frame relay trial. The objective is to demonstrate
the interoperability of frame relay technology in the Open System
Interconnection (OSI) environment. Frame relay is an emerging technology
which provides interconnection, both directly and indirectly, among
subnetworks in OSI environment.
About 80 vendors came together to form the Fast Ethernet Alliance (FEA). The
goal of the FEA was to speed Fast Ethernet through the Institute of Electrical
and Electronic Engineers (IEEE) 802.3 body, the committee that controls the
standards for Ethernet. The FEA succeeded in June 1995, and the technology
passed a full review and was formally assigned the name 802.3u. The IEEE's
name for Fast Ethernet is OOBaseT. 100BaseTis an extension of the 10BaseT
standard, designed to raise the data transmission capacity of 10OBaseT from
10Mbps to 100Mbps.
100 VG-AnyLAN is a new IEEE 802.12 technology for transmitting Ethernet
(802.3) and token-ring (802.5) frame information at 100M bits per second. The
actual standard was jointly developed by IBM Corp, AT&T Corp and Hewlett Packard Co. A group of 13 computer networking companies have agreed on the
way forward for high- speed data networking and announced products based on
a new technology standard known as 100 VG-AnyLAN. This coalition is
competing with the FEA in designing a standard that allows existing networks
to be upgraded to move data at a rate of 100Mbit/s
About 12 firms formed a consortium to develop interoperable products
incorporating the Full Duplex Switched Ethernet access method. Full-duplex
Ethernet boosts speeds by allowing data to be transferred in both directions at
once on a dedicated Ethernet segment, something that's usually prevented by
CSMA/CD.
25 vendors formed the ATM25 Alliance to accelerate the acceptance of
Asynchronous Transfer Mode (ATM) at the desktop. The group's charter is to
make 25Mbps ATM products open, accessible and interoperable with other
leading ATM products.
16 vendors of Token-Ring networking products formed an alliance to promote
their ties with customers, assuring them that new products are designed to be
interoperable.
Apple Computer, AT&T, IBM and Siemens formed Versit, a global initiative to
enable interoperability between existing and emerging communication and
information products. The cooperation covers products including telephones,
PBXs, computers, networks, servers and personal digital assistants.
Four companies formed the ISDN Forum to simplify interoperability between
ISDN customer premise equipment (CPE) and the public switched network.
3Com, Ascend Communications, AT&T Network Systems, and U.S. Robotics
are the forum's founders.
67
Table F-1.
Interoperability Forums and Standards-setting Alliances,
Continued
Name of Alliance and Date
of Formation*
11. ATM RMON
February 1996
12. Wireless LAN
April 1996
Objective of Alliance
Seven firms collaborate to develop specification for ATM remote monitoring
and analysis tool solutions to advance the management of ATM networks.
13 vendors of wireless LAN formed alliance to promote awareness of wireless
LAN. It does not plan to get involved in standards making or interoperability;
instead it creates an advisory committee for the industry, composed of
customers, independent software vendors and systems integrators.
More than 50 firms have joined the Gigabit Ethernet Alliance to develop
13. Gigabit Ethernet
Gigabit Ethernet Standard that allows half and full duplex operations at 1,000
Alliance
May 1996
megabits per second. In addition, it also addresses backward compatibility with
10BaseT and 100BaseT technologies. The Gigabit task force will actively
pursue specification development of protocol, media access controller, repeater
and physical layer technologies.
13 ISDN customer premise equipment (CPE) manufacturers have formed a new
14. Vendor's ISDN
Association
organization -- called Vendors' ISDN Association (VIA), previously was known
as the ISDN Forum. The group has unified CPE vendors on key market and
June 1996
technical issues that will help the entire ISDN industry grow.
*Date of formation is approximate only.
Source: Lexis/Nexis: CMPCOM database, 1996.
68
References
Abernathy, W., & Clark, K., (1985), "Innovation: Mapping the winds of creative
destruction," Research Policy, 14(1), 3-22.
Abernathy, W., & Utterback, J., (1978), "Patterns of industrial innovation," Technology
Review, 41-47.
Abrahamson, E., & Rosenkopf, L., (1997), "Social network effects on the extent of
innovation diffusion: A computer simulation," OrganizationScience, 8(3), 289-309.
Aldrich, H., & Whetten D. A., (1981), "Organization-sets, action-sets, and networks:
making the most of simplicity." In P. C. Nystrom & W. H. Starbuck, Handbook of
OrganizationDesign. New York, Oxford University Press. 1: 385-408.
Allen, T., J., (1977), Managing the Flow of Technology. Cambridge: Mass: MIT Press.
Arora, A. & Gambardella A., (1995), "The division of innovative labor in biotechnology.
Sources of Medical Technology: Universities and Industry." In N. Rosenberg, A. C.
Gelijns & H. Dawkins. Washington, D.C., National Academy Press: 188-205.
Axelrod, R., Mitchell, W., Thomas, R. E., Bennett, S. D., & Bruderer, E., (1995),
"Coalition formation in standard-setting alliances," Management Science, 41(9), 14931508.
Barley, S. R., Freeman, J., & Hybels, R., (1992), "Strategic alliances in commercial
biotechnology." In N. Nohria & R. Eccles (Eds.), Networks and Organizations (311-347).
Boston, MA: Harvard Business School Press.
Barney, J. B., (1991), "Firm resources and sustained competitive advantage," Journal of
Management, 17, 99-120.
Baum, J., & Singh, J., (1994), "Organization-environment coevolution." In J. Baum & J.
Singh (Eds.), Evolutionary Dynamics of Organizations. New York: Oxford University
Press.
Baum, J., & Oliver, C., (1991), "Institutional linkages and organizational mortality,"
Administratively Science Quarterly, 36, 187-218.
Bavelas, A., (1948), "A mathematical model for group structures," Human Organization,
7, 16-30.
69
Bavelas, A., & Barrett, D., (1951), "An experimental approach to organizational
communication," Personnel,27, 366-371.
Bradach, J. L., & Eccles, R. G., (1989), "Price, authority, and trust: From ideal types to
plural forms," Annual Review of Sociology, 15, 97-118.
Burt, R., (1980), "Innovation as a structural interest: Rethinking the impact of network
position on innovation adoption," Social Networks, 2, 327-355.
Burt, R., (1992a), "The social structure of competition." In N. Nohria & R. Eccles (Eds.),
Networks and Organizations(57-91). Boston, MA: Harvard Business School Press.
Burt, R., (1992b), Structural Holes: The Social Structure of Competition. Cambridge,
MA: Harvard University Press.
Christensen, C., (1992a), "Exploring the limits of the technology S-curve. Part I:
Component technologies," Productionand OperationsManagement, 1(4), 334-357.
Christensen, C., (1992b), "Exploring the limits of the technology S-curve. Part II:
Architectural technologies," Productionand OperationsManagement, 1(4), 358-366.
Christensen, C., & Rosenbloom, R., (1995), "Explaining the attacker's advantage:
Technological paradigms, organizational dynamics, and the value network," Research
Policy, 24, 233-257.
Clarysse, B., Debackere, K., & Dierdonck, R. V., (1996), "Research networks and
organizational mobility in an emerging technological field: The case of plant
biotechnology," Economics of Innovation and New Technology, 4, 77-96
Coleman, J. (1990), Foundations of Social Theory, (Chapter 12, 300-321). Cambridge,
MA: Harvard University Press.
Coleman, J., Katz, E., & Menzel, H., (1957), "The diffusion of an innovation among
physicians," Sociometry, 20, 253-270.
Contractor, F. J., & Lorange, P., (1988), "Why should firms cooperate? The strategy and
economics basis for cooperative venture." In F. J. Contractor & P. Lorange (Eds.),
Cooperative Strategies in International Business (3-30). Lexington, MA: Lexington
Books.
Coxon, A. P. M., (1982), The User's Guide to Multidimensional Scaling. Great Britain:
Heinemann Educational Books.
70
Czepiel, J. A., (1975), "Patterns of interorganizational communications and the diffusion
of a major technological innovation in a competitive industrial community," Academy of
Management Journal, 18(1), 6-24.
DeBresson, C., & Amesse, F., (1991), "Networks of innovators: A review and
introduction to the issue," Research Policy, 20, 363-379.
Dimaggio, P., (1986), "Structural analysis of organizational fields: A blockmodel
approach," Research in OrganizationalBehavior, 8, 335-370.
DiMaggio, P., (1994), "Commentary: The challenge of community evolution." In J.
Baum & J. Singh (Eds.), EvolutionaryDynamics of Organizations(444-450). New York:
Oxford University Press.
Fombrun, C. J., (1982), "Strategies for network research in organizations," Academy of
Management Review, 7(2), 280-291.
Freeman, C., (1991), "Networks of innovators: A synthesis of research issues," Research
Policy, 20, 499-514.
Freeman, C., (1994), "Critical survey," CambridgeJournalof Economics, 18, 463-514.
Freeman, L. C., (1978), "Centrality in social networks: Conceptual clarification," Social
Networks, 1, 215-239.
Garud, R., & Kumaraswamy, A., (1993), "Changing competitive dynamics in network
industries: An exploration of Sun Microsystems' open systems strategy," Strategic
Management Journal, 14, 351-369.
Garrette, B., & Dussauge, P., (1995), "Patterns of strategic alliances between rival firms,"
Group Decision and Negotiation, 4, 429-452.
Granovetter, M., (1992), "Problems of explanation in economic sociology." In N. Nohria
& R. Eccles (Eds.), Networks and Organizations: Structure, Form and Action (25-56).
Boston, MA: Harvard Business School Press.
Gray, B., (1990), "Building interorganizational alliances: Planned change in a global
environment," Research in OrganizationalChange and Development, 4, 101-140.
Gulati, R., (1995), "Social structure and alliance formation patterns: A longitudinal
analysis," Administrative Science Quarterly, 40, 619-652.
71
Hagedoorn, J., (1993), "Understanding the rationale of strategic technology partnering:
Interorganizational modes of cooperation and sectoral differences," Strategic
Management Journal, 14, 371-385.
Hagedoorn, J., (1995), "Strategic technology partnering during the 1980s: Trends,
networks and corporate patterns in non-core technologies," Research Policy, 24, 207-231.
Hakansson, H., & Johanson, J., (1988), "Formal and informal cooperation strategies in
international industrial networks." In F. J. Contractor & P. Lorange (Eds.), Cooperative
Strategies in InternationalBusiness (369-379). Lexington, MA: Lexington Books.
Hamel, G., (1991), "Competition for competence and inter-partner learning within
international strategic alliances," Strategic ManagementJournal, 12, 83-103.
Harrigan, K. R., (1987), "Strategic alliances: Their new role in global competition,"
Columbia Journal of World Business, Summer, 67-69.
Henderson, R., & Clark, K. B., (1990), "Architectural innovation: The reconfiguration of
existing product technologies and the failure of established firms," Administrative
Science Quarterly, 35, 9-30.
Hergert, M., & Morris, D., (1988), "Trends in international collaborative agreements." In
F. Contractor & P. Lorange (Eds.), CooperativeStrategies in InternationalBusiness (99109). Lexington, MA: Lexington Books.
Hughes, T. P., (1987), "The evolution of large technological systems." In W. E. Bijker, T.
P. Hughes, & T. Pinch (Eds.), The Social Construction of Technological Systems (51-82).
Cambridge, MA: MIT Press.
Jones, C., Hesterly, W. S., & Borgatti, S. P., (1997), "A general theory of network
governance: Exchange conditions and social mechanisms," Academy of Management
Review, 22(4), 911-945.
Kogut, B., & Zander, U., (1992), "Knowledge of the firm, combinative capabilities, and
the replication of technology," OrganizationScience, 3(3), 383-397.
Liebeskind, J. P., Oliver, A. L., Zucker, L., & Brewer, M., (1996), "Social networks,
learning and flexibility: Sourcing scientific knowledge in new biotechnology firms,"
OrganizationScience, 7(4), 428-443.
Mariti, P., & Smiley, R. H., (1983), "Cooperative agreements and the organizational of
industry," The Journalof IndustrialEconomics, 31(4), 437-451.
72
Marsden, P. V., (1990), "Network data and measurement," Annual Review of Sociology,
16, 435-463.
Miller, R., Hobday, M., Leroux-Demers, T., & Olleros, X., (1995), "Innovation in
complex systems industries: The case of flight simulaton," Industrial and Corporate
Change, 4(2), 363-400.
Mitchell, W., & Singh, K., (1992), "Incumbents' use of pre-entry alliances before
expansion into new technical subfields of an industry," Journal of Economic Behavior
and Organization, 18, 347-372.
Mizruchi, M., & Galaskiewicz, J., (1993), "Networks of interorganizational relations,"
SociologicalMethods and Research, 22(1), 46-70.
Morgan, D. L., Neal, M. B., & Carder, P., (1996), "The stability of core and peripheral
networks over time," Social Networks, 19, 9-25.
Nohria, N., & Eccles, R., (1992), Networks and Organizations: Structure, Form and
Action. Boston, MA: Harvard University Press.
Nohria, N. & Garcia-Pont C., (1991), "Global strategic linkages and industry structure,"
Strategic Management Journal, 12: 105-124.
Olleros, F.J., & Macdonald, R. J., (1988), "Strategic alliances: Managing
complementarity to capitalize on emerging technologies," Technovation, 7, 155-176.
Ouchi, W. G., (1980), "Markets, bureaucracies, and clans," Administrative Science
Quarterly, 25, 129-142.
Powell, W., Koput, K., & Laurel, S., (1996), "Interorganizational collaboration and the
locus of innovation: Networks of learning in biotechnology," Administrative Science
Quarterly, 41(1), 116-14
Richardson, G. B., (1972), "The organization of industry," Economic Journal, 82, 883896.
Ring, P. S., & Van de Ven, A. H., (1992), "Structuring cooperative relationships between
organizations," Strategic Management Journal, 13, 483-498.
Roberts, E. B., & Berry, C. A., (1985), "Entering new businesses: Selecting strategies for
success," Sloan Management Review, 3-19.
73
Rosenbloom, R. S., & Cusumano, M. A., (1987), "Technological pioneering and
competitive advantage: The birth of the VCR industry," California Management Review,
29(4), 51-76.
Rosenkopf, L., & Tushman, M., (1994), "The coevolution of technology and
organization." In J. Baum & J. Singh (Eds.), Evolutionary Dynamics of Organizations
(403-424). New York: Oxford University Press.
Schiffman, S., Reynolds, M., & Young, F., (1981), Introduction to Multidimensional
ScaliTheory, Methods, and Applications. New York: Academic Press.
Schumpeter, J. A. (1942). Capitalism, Socialism and Democracy. New York, Harper &
Brothers.
Sinha, D. K., & Cusumano, M., (1991), "Complementary resources and cooperative
research: A model of research joint ventures among competitors," Management Science,
37(9), 1091-1106.
Stork, D., (1991), "A longitudinal study of communication networks: Emergence and
evolution in a new research organization," Journal of Engineering and Technology
Management, 7, 177-196.
Teece, D., (1986), "Profiting from technological innovation: Implications for integration,
collaboration, licensing and public policy," Research Policy, 15, 285-305.
Teece, D., (1992), "Strategies for capturing the financial benefits from technological
innovation." In N. Rosenberg, R. Landau, & D. Mowery (Eds.), Technology and the
Wealth of Nations (175-205). Stanford, California: Stanford University Press.
Thorelli, H. B., (1986), "Networks: between markets and hierarchies," Strategic
Management Journal,7, 37-51.
Tichy, N. M., Tushman, M. L., & Fombrun, C., (1979), "Social network analysis for
organizations," Academy of Management Review, 4(4), 507-519.
Tucci, C., & Cusumano, M.,(1994), Benefits and pitfalls of international strategic
technology alliances (Working Paper No. 110-94): International Center for Research on
the Management of Technology, Massachusetts Institute of Technology.
Tushman, M., & Anderson, P., (1986), "Technological discontinuities and organizational
environment," Administrative Science Quarterly, 31, 439-456.
74
Tushman, M. L., & Rosenkopf, L., (1992), "Organizational determinants of technological
change: Toward a sociology of technological evolution," Research in Organizational
Behavior, 14, 311-347.
Utterback, J., & Suarez, F., (1993), "Innovation, competition, and industry structure,"
Research Policy, 22, 1-21.
Van de Ven, A., & Garud, R., (1994), "The coevolution of technical and institutional
events in the development of an innovation." In J. Baum & J. Singh (Eds.), Evolutionary
Dynamics of Organizations.New York: Oxford University Press.
Walker, G., Kogut, B., & Shan, W., (1997), "Social capital, structural holes and the
formation of an industry network," OrganizationScience, 8(2), 109-125.
Wasserman, S., & Faust, K., (1994), Social Network Analysis: Methods and Applications.
New York: Cambridge University Press.
Wernerfelt, B., (1984), "A resource-based view of the firm," Strategic Management
Journal, 5, 171-180.
Williamson, O. E., (1985), The Economic Institutions of Capitalism: Firms, Markets,
Relational Contracting.New York: Free Press.
Yan, A., & Gray, B., (1994), "Bargaining power, management control, and performance
in United States-China joint ventures: A comparative case study," Academy of
Management Journal, 37(6), 1478-1517.
75